Konnichiwa,
And? The one thing that would worry me a little is the midrange bump between around 400-900Hz and the peak in the treble around 10KHz, BUT these peaks are non too bad AND fall each well within one of the Blauert bands that control percieved directionality and diffuse/direct perception and hence they are likely to enhance the perception of 3-dimensionality.
Given that 3A's designer designs "by ear" and seems to prefer a D to an A (as in 3D instead of 3A) it is pleasing to find basic acoustic theory validated.
If THAT looks awfull (, have you actually ever taken measurements at the listening position in your living/listening room?
Apart from the two boost region which I suspect are the main reason these speakers sound so "3D" the response is essentially flat (check against stereophile measurements and observe scales on the curves) and hence, apart from what appears some deliberate tailoring of the sound with a certain design goal the response is pretty even, certainly even enough that the frequency response at any listening position is determined almost exclusively by room effects and positioning.
Yes.
Sayonara
morbo said:
And? The one thing that would worry me a little is the midrange bump between around 400-900Hz and the peak in the treble around 10KHz, BUT these peaks are non too bad AND fall each well within one of the Blauert bands that control percieved directionality and diffuse/direct perception and hence they are likely to enhance the perception of 3-dimensionality.
Given that 3A's designer designs "by ear" and seems to prefer a D to an A (as in 3D instead of 3A) it is pleasing to find basic acoustic theory validated.
morbo said:
If THAT looks awfull (, have you actually ever taken measurements at the listening position in your living/listening room?
Apart from the two boost region which I suspect are the main reason these speakers sound so "3D" the response is essentially flat (check against stereophile measurements and observe scales on the curves) and hence, apart from what appears some deliberate tailoring of the sound with a certain design goal the response is pretty even, certainly even enough that the frequency response at any listening position is determined almost exclusively by room effects and positioning.
morbo said:
Yes.
Sayonara
Designs like the 3A MM DeCapo which allow a wide/full-range driver to be directly connected to the driving amp have a very live and immediate sound which many find highly attractive. I would postulate two reasons:
1. No filters between you and the signal. OK, so it is actually one less (very lossy) filter in a long chain of filters the typical signal is subjected to between original event and reproduction in my living room. The point is the mid-woofer is connected directly to the amp. Talk to that all too rare breed of audiophile who has gone ‘true’ bi-amp and inevitably they speak of huge improvements in transparency along with the many other benefits. In this context by “true bi-amp” I mean the passive high level filters in the speaker box are GONE and the crossover is a low level device upstream of the amplifiers. I find it ironic that as audiophiles we obsess about a few feet of speaker cable when typically as soon as that signal gets in the speaker box it hits 50 feet of 20awg magnet wire coiled up in inductor(s). The terminally tweaked among us will upgrade the inductors to big air cores or foil wire types. Perhaps the ultimate upgrade is to get rid of the inductor all together?
2. Better time/phase behavior. Get rid of the typical crossover LPF (low pass filter) and you get rid of its phase shift. My experience is some of us seem to be quite sensitive to non-linear phase speakers and find ourselves drawn to more phase correct designs like full range electrostats, full range single driver types (Lowther, Fostex, etc.) and the likes of Thiel, Vandersteen, and B&W who pay attention to the time as well as frequency domains. Other listeners have differing musical priorities and are perfectly happy with speakers designed with no regard for the time domain. Now the 3A MM DeCapo design being discussed will have phase shift as a result of the filter action when its mid-woofer acts as a mechanical LPF. But does not the typical driver behave in the same fashion to which is added another set of phase shifts thanks to the caps and inductors in its crossover? Now of course for years now clever designers (Richard Vandersteen and Jim Thiel come immediately to mind) have taken all this into account, lining up all the phase shifts resulting from the drivers and crossovers being bandpass devices to produce “linear phase” speakers. Maybe just chopping out the LPF on the 3A MM DeCapo gets many of the same benefits in a much simpler configuration?
1. No filters between you and the signal. OK, so it is actually one less (very lossy) filter in a long chain of filters the typical signal is subjected to between original event and reproduction in my living room. The point is the mid-woofer is connected directly to the amp. Talk to that all too rare breed of audiophile who has gone ‘true’ bi-amp and inevitably they speak of huge improvements in transparency along with the many other benefits. In this context by “true bi-amp” I mean the passive high level filters in the speaker box are GONE and the crossover is a low level device upstream of the amplifiers. I find it ironic that as audiophiles we obsess about a few feet of speaker cable when typically as soon as that signal gets in the speaker box it hits 50 feet of 20awg magnet wire coiled up in inductor(s). The terminally tweaked among us will upgrade the inductors to big air cores or foil wire types. Perhaps the ultimate upgrade is to get rid of the inductor all together?
2. Better time/phase behavior. Get rid of the typical crossover LPF (low pass filter) and you get rid of its phase shift. My experience is some of us seem to be quite sensitive to non-linear phase speakers and find ourselves drawn to more phase correct designs like full range electrostats, full range single driver types (Lowther, Fostex, etc.) and the likes of Thiel, Vandersteen, and B&W who pay attention to the time as well as frequency domains. Other listeners have differing musical priorities and are perfectly happy with speakers designed with no regard for the time domain. Now the 3A MM DeCapo design being discussed will have phase shift as a result of the filter action when its mid-woofer acts as a mechanical LPF. But does not the typical driver behave in the same fashion to which is added another set of phase shifts thanks to the caps and inductors in its crossover? Now of course for years now clever designers (Richard Vandersteen and Jim Thiel come immediately to mind) have taken all this into account, lining up all the phase shifts resulting from the drivers and crossovers being bandpass devices to produce “linear phase” speakers. Maybe just chopping out the LPF on the 3A MM DeCapo gets many of the same benefits in a much simpler configuration?
Meaningful Measures
John - I read that paper - and its interesting research no doubt.
However there's a few things that make me yawn.
1) We know people like harmonic distortion as long as its the right kind. Take it from a guitar player. So that's not a huge surprise but still interesting.
2) The levels on IM distortion in the samples would no doubt sound like hell - no surprise there either.
3) I can't help but wonder if he played the distortion on more accurate speakers if it would have sounded worse. 🙂
The question I'm asking is, in effect, What composite of attributes looks like the right balance? Can we make a formula that takes phase, driver agility/compliance/speed, FR, comb filering, distortion of different types, sound stage (which might be a function of directivity and dispersion), and other attributes, and roughly weight them such that a high number is a speaker that sounds better? I'm looking for the variables and the coeffecients.
John - I read that paper - and its interesting research no doubt.
However there's a few things that make me yawn.
1) We know people like harmonic distortion as long as its the right kind. Take it from a guitar player. So that's not a huge surprise but still interesting.
2) The levels on IM distortion in the samples would no doubt sound like hell - no surprise there either.
3) I can't help but wonder if he played the distortion on more accurate speakers if it would have sounded worse. 🙂
The question I'm asking is, in effect, What composite of attributes looks like the right balance? Can we make a formula that takes phase, driver agility/compliance/speed, FR, comb filering, distortion of different types, sound stage (which might be a function of directivity and dispersion), and other attributes, and roughly weight them such that a high number is a speaker that sounds better? I'm looking for the variables and the coeffecients.
I think that's the point where personal taste and differing accuracy of listeners come into play. Therefore it will be almost impossible to determine these coefficients.
Regards
Charles
I know that Mr. Linkwitz places a great deal of emphasis on a flat frequency response and coherent phase response.
From his theory on speakers, a fun speaker is inaccurate and deviates from the recording itself. If the recording sounds fun then so does the speaker. If the recording is boring then so is the speaker. This is how a speaker should act. A speaker that always sounds boring has some design issues and a speaker that always sounds fun also has some design issues. Namely in the FR response department.
I find that at live performances the cymbals sound with greater volume than the rest of the band, my goal is to recreate this with an slowly raised top-end, but not for a broad area. It is tough and I've come very close to the sound that I remember. Of course, balance with midrange is more important than the raise in the higher frequencies. I'm tapping my feet and the speakers aren't in an enclosure yet.
From his theory on speakers, a fun speaker is inaccurate and deviates from the recording itself. If the recording sounds fun then so does the speaker. If the recording is boring then so is the speaker. This is how a speaker should act. A speaker that always sounds boring has some design issues and a speaker that always sounds fun also has some design issues. Namely in the FR response department.
I find that at live performances the cymbals sound with greater volume than the rest of the band, my goal is to recreate this with an slowly raised top-end, but not for a broad area. It is tough and I've come very close to the sound that I remember. Of course, balance with midrange is more important than the raise in the higher frequencies. I'm tapping my feet and the speakers aren't in an enclosure yet.
Attachments
Bose(o) said:
With all due respect to Mr Linkwitz' technical nous, his opinion on what constitutes fun listening is no more valid than yours or mine. You will never even get close to the original recording, (assuming it was recorded in a 'purist' manner, and if it wasn't, in trying to, aren't you chasing a phantom?) so why not sit back and enjoy the ride? If listening to a speaker with a FR that resembles a 'smiley face' on a 31 band EQ makes me want to get up and dance, sing, groove or get whatever sort of enjoyment it is I get out of most of my recordings, then it's a GREAT speaker.
Power response and distortion are the issues that SL put highest at his list. His speakers has gross phase distortion as most others.
/Peter
/Peter
After listening to my sister play guitar, I figured that a flat response (not being obsessive about flatness of course as in my FR graph) yet, allowing the enclosure and such to resonate as any other instrument would, will yield the best sound. The enclosure resonance could be controlled to stay within some acceptable FR response as to keep other sounds in balance.
I am enjoying this thread very much, thank you for speaking your opinions as I am in the measurement/design phase of my next gen of speakers! What is learned from this thread will probably be applied in the design.
I am enjoying this thread very much, thank you for speaking your opinions as I am in the measurement/design phase of my next gen of speakers! What is learned from this thread will probably be applied in the design.
Norman Tracy said:
Norm has pretty much already posted what i was going to say...
I have heard these speakers, and for commercial jobs they are very good. and deserve the accolades. Their price is just the reality of the commercial speaker market.
A speaker like this with very low reactance can sound much more dynamic & will inherently have better time response than a speaker with a more complex XO -- and since the human ear/brain is so sensitive to time response that can overwelm deviations in the amplitude response.
As DIYers there are a large number of drivers (and gowing) that can be used for a system of similar concept.
dave
planet10/big fan of the (mostly) full-range speaker concept.
"allowing the enclosure and such to resonate as any other instrument would, "
But a speaker is not meant to be a instrument, rather it´s meant to be a transducer, at least that´s the way I see it. A speaker with gross resonances will color the music with the same palette to every tune, masking the real sound of the record.
The different resonant behaviours of acoustic instruments is part of what makes them sound the way they do, and to let that shine thru you need a speaker as neutral as can be.
/Peter
But a speaker is not meant to be a instrument, rather it´s meant to be a transducer, at least that´s the way I see it. A speaker with gross resonances will color the music with the same palette to every tune, masking the real sound of the record.
The different resonant behaviours of acoustic instruments is part of what makes them sound the way they do, and to let that shine thru you need a speaker as neutral as can be.
/Peter
I too have extensive experience with the Reference 3A designs. These were used for years by a very well known mastering engineer whom I worked for. They are very revealing and VERY dynamic. The designer spent years devloping the woofer so that is was truely capable of running with no XO. Expensive by DIY standards for a simple 2 way, but very much worth the price in terms of performance -vs- market hi-fi prices.
Greg
Greg
Pan said:
A neutral speaker being a boring speaker as others have put forth.
I don't agree with that because I come home after school or work anxiously awaiting to play the hymns of Hawksley Workman. The detail and breathness of OBs are so intriguing and very engaging. I still believe that a loudspeaker shall rely on the recording for its sound.
"A neutral speaker being a boring speaker as others have put forth. "
If a neutral speaker sounds boring, then 1- it´s actually not neutral or 2- someting major wrong in the chain.
A speaker that is not neutral will have some coloration, and that coloration will add a "sameness" (to paraphrase SL) to all recordings... if that is not boring I don´t know what. A low distortion neutral speaker have the "power" to do whatever it is told to do... that is fun!
/Peter
If a neutral speaker sounds boring, then 1- it´s actually not neutral or 2- someting major wrong in the chain.
A speaker that is not neutral will have some coloration, and that coloration will add a "sameness" (to paraphrase SL) to all recordings... if that is not boring I don´t know what. A low distortion neutral speaker have the "power" to do whatever it is told to do... that is fun!
/Peter
Generally, I prefer speakers with flat frequency response.
In the 80's when car audio was shockingly expensive, I noted that $15/pair paper full range drivers sounded considerably better than $60 coaxial two-ways. ($60 back then was more like $120 now) It wasn't until you got into the near $100 price that that multi-way speakers did better overall than the cheap speakers. Perhaps the more expensive speakers were also doing a better job with the crossover, too, but this could have also been just the more dynamic sound of the full-range driver.
My current speakers, a GR-Research-T1 kit do a rather nice job with a fairly simple crossover, yet they are both time-aligned and reasonably flat in response. They don't use a mechanical crossover of the woofer, however.
Part of the reason is the drivers were designed to have flat response in the first place so that they are easy to cross over without having to add extra components to smooth out resonance peaks in the tweeter or the woofer.
I had independent confirmation of the superiority of frequency response several times when I discovered that various speakers that I had picked out of others as being better were confirmed by magazine tests to have a good frequency response curve. I did not know any of this before hand.
I really don't understand why designing a speaker with a mechanical crossover is any better than doing it with capacitors and coils. There are going to be phase shifts and distortion either way.
I also don't understand speaker designers who don't measure their speakers. Measuring frequency response can help you track down colorations in the sound that would take a lot of listening to fix. Knowing the exact frequency of the coloration makes it easier to make a small crossover mod that takes care of the problem.
I also don't think you should let too much faith in numbers lead you astray, either. Hearing is subjective, so if you get too worried about what you can measure, you will probably compromise something that is harder to measure, but you can still hear.
In the 80's when car audio was shockingly expensive, I noted that $15/pair paper full range drivers sounded considerably better than $60 coaxial two-ways. ($60 back then was more like $120 now) It wasn't until you got into the near $100 price that that multi-way speakers did better overall than the cheap speakers. Perhaps the more expensive speakers were also doing a better job with the crossover, too, but this could have also been just the more dynamic sound of the full-range driver.
My current speakers, a GR-Research-T1 kit do a rather nice job with a fairly simple crossover, yet they are both time-aligned and reasonably flat in response. They don't use a mechanical crossover of the woofer, however.
Part of the reason is the drivers were designed to have flat response in the first place so that they are easy to cross over without having to add extra components to smooth out resonance peaks in the tweeter or the woofer.
I had independent confirmation of the superiority of frequency response several times when I discovered that various speakers that I had picked out of others as being better were confirmed by magazine tests to have a good frequency response curve. I did not know any of this before hand.
I really don't understand why designing a speaker with a mechanical crossover is any better than doing it with capacitors and coils. There are going to be phase shifts and distortion either way.
I also don't understand speaker designers who don't measure their speakers. Measuring frequency response can help you track down colorations in the sound that would take a lot of listening to fix. Knowing the exact frequency of the coloration makes it easier to make a small crossover mod that takes care of the problem.
I also don't think you should let too much faith in numbers lead you astray, either. Hearing is subjective, so if you get too worried about what you can measure, you will probably compromise something that is harder to measure, but you can still hear.
I think non-neutral has issues because...
If a bit of program material happens to lean on the place where the coloration lives, it sounds terrible. For example, a lot of people like a little extra high end. That sounds great on lots of stuff but on records with any hint of harshness its bad.
I'd say coloration can compliment most source material, but its different for different material. Ergo if you listen to the same sort of stuff, but all means have coloration that compliments it - but if you want a general reference you are ill advised to favor coloration. Another example is, in traditional Jazz, the Bass usually has too much "fuff" - around 100 hz. If you have a lot of coloration there, or in some spot like 110-140, the low end in a lot of Jazz recordings will sound bad. I have this issue recording string basses. EQ out some of that 100hz stuff, or leave it in and let those with bassy speakers suffer.
If a bit of program material happens to lean on the place where the coloration lives, it sounds terrible. For example, a lot of people like a little extra high end. That sounds great on lots of stuff but on records with any hint of harshness its bad.
I'd say coloration can compliment most source material, but its different for different material. Ergo if you listen to the same sort of stuff, but all means have coloration that compliments it - but if you want a general reference you are ill advised to favor coloration. Another example is, in traditional Jazz, the Bass usually has too much "fuff" - around 100 hz. If you have a lot of coloration there, or in some spot like 110-140, the low end in a lot of Jazz recordings will sound bad. I have this issue recording string basses. EQ out some of that 100hz stuff, or leave it in and let those with bassy speakers suffer.
I don't believe that a mechanical crossover per se is superior to an electrical one either.
But keep in mind that every driver has a mechanical rolloff at its top end anyway. So using an electrical 2nd order filter with your woofer will result in in a fourth-order lowpass if the driver shows a natual 2nd order lowpass behaviour.
If you can design your driver to behave like your target 2nd order lowpass you'll end up with better time-alignment.
Regards
Charles
Pan said:
Heard that. Saying a neutral speaker is boring is like saying a well-mannered girlfriend is boring because there's no drama. It starts getting tired around the third time the crockery flies, and for speakers it gets tired when the violas start running into the cellos.
Francois.
Most speaker designs don't attempt this mechanical crossout for very good technical reasons.
For one, it is really hard to make the rolloff smooth. Not only that, most woofers will have a peak or two in their output in the range that you are crossing them out at. There are also cone-breakup issues.
With a crossover and carefully selecting values you can solve many of these problems very easily. You don't have to care about cone breakup, because you can cross out before it starts to occur.
When you eliminate the crossover, you have to solve these problems some other way and it almost certainly means some other tradeoff.
For example to tame cone breakup, you probably have to make the cone a bit more rigid than what would be ideal, so the woofer has a rise in output as the frequency goes up, since the higher frequencies aren't being decoupled from the outer edges of the driver in a way that leads to more linear response. You may also have to add mass to the cone to get this stiffness, so the speaker is less efficient.
And since you don't have a crossover, you can't tame this midrange rise either. You have to live with it.
The idea that eliminating capacitors and coils from a woofers somehow makes it more dynamic doesn't make sense, either. It is very likely that added mass to the center of the cone is one way that is used to get the mechanical crossout. This extra mass means that it takes more motor energy to move the speaker, so it is less efficient. This means that the speaker is less dynamic, not more.
All of this adds up to mean that a speaker that tries to mechanically cross out the woofer almost always does so at the expense of frequency response, for a supposed gain in accuracy due to leaving out a the crossover components.
I have seen aluminum cone speakers touting an "audiophile" mechanical crossover of the woofer and this seems astonshing since aluminum cones are rather infamous for their problems with upper-frequency cone breakup. I didn't get a chance to hear how they sounded since they were in a catalog. But I doubt that it would work very well.
The speaker that is the subject of this thread seems to have some pretty ragged frequency response. The mechanical rolloff is not particularly even. I also doubt that that the phase is consistent in the rolloff, due to things like cone breakup in the upper end of the rolloff.
Cone breakup results in a very distorted sound as well, because it produces ringing artifacts in the sound.
I seriously doubt that there is anything it does better that compensates for the poor frequency response results. Some people may focus on a few things that it does well, and not hear the other problems though.
On some material you may find that the peaks and dips don't bother you much, but I suspect that if you had to live with them for a long time you would find a bunch of recordings where the peaks and dips were rather annoying with these speakers.
Some designers do acheive a fairly good degree of phase alignment with capacitors and coils in the signal path, and they can do this without so many compromises in frequency response, which is not the sole critieria for good sound, but it is rather important.
For one, it is really hard to make the rolloff smooth. Not only that, most woofers will have a peak or two in their output in the range that you are crossing them out at. There are also cone-breakup issues.
With a crossover and carefully selecting values you can solve many of these problems very easily. You don't have to care about cone breakup, because you can cross out before it starts to occur.
When you eliminate the crossover, you have to solve these problems some other way and it almost certainly means some other tradeoff.
For example to tame cone breakup, you probably have to make the cone a bit more rigid than what would be ideal, so the woofer has a rise in output as the frequency goes up, since the higher frequencies aren't being decoupled from the outer edges of the driver in a way that leads to more linear response. You may also have to add mass to the cone to get this stiffness, so the speaker is less efficient.
And since you don't have a crossover, you can't tame this midrange rise either. You have to live with it.
The idea that eliminating capacitors and coils from a woofers somehow makes it more dynamic doesn't make sense, either. It is very likely that added mass to the center of the cone is one way that is used to get the mechanical crossout. This extra mass means that it takes more motor energy to move the speaker, so it is less efficient. This means that the speaker is less dynamic, not more.
All of this adds up to mean that a speaker that tries to mechanically cross out the woofer almost always does so at the expense of frequency response, for a supposed gain in accuracy due to leaving out a the crossover components.
I have seen aluminum cone speakers touting an "audiophile" mechanical crossover of the woofer and this seems astonshing since aluminum cones are rather infamous for their problems with upper-frequency cone breakup. I didn't get a chance to hear how they sounded since they were in a catalog. But I doubt that it would work very well.
The speaker that is the subject of this thread seems to have some pretty ragged frequency response. The mechanical rolloff is not particularly even. I also doubt that that the phase is consistent in the rolloff, due to things like cone breakup in the upper end of the rolloff.
Cone breakup results in a very distorted sound as well, because it produces ringing artifacts in the sound.
I seriously doubt that there is anything it does better that compensates for the poor frequency response results. Some people may focus on a few things that it does well, and not hear the other problems though.
On some material you may find that the peaks and dips don't bother you much, but I suspect that if you had to live with them for a long time you would find a bunch of recordings where the peaks and dips were rather annoying with these speakers.
Some designers do acheive a fairly good degree of phase alignment with capacitors and coils in the signal path, and they can do this without so many compromises in frequency response, which is not the sole critieria for good sound, but it is rather important.
I like to associate beaming with cone breakup since, in the breakup regions the speaker usually doesn't hold well for off-axis response.
I also think that these work together to give you the harsh sound that most drivers produce without an inductor. Now, whether or not the roll-off was smooth my ears could detect that sound of beaming...narrowness if you will, in which the voices made it seem like there were little people singing on the cone.
Even if the roll-off is smooth enough for the KISS principle, I found that the sound was cleaner and more relaxed and refined with at least an inductor in place.
Oh, btw, I generally use 1st-order crossovers on my designs. This is to keep the character of the speakers or the culmination of materials and design relevant to the overall sound of the speaker. i.e., why go carbon fibre without getting a hint of the sound of carbon fibre.
Of course I do this within reason and would work on getting rid of large peaks especially when they are associated with stored energy.
I also think that these work together to give you the harsh sound that most drivers produce without an inductor. Now, whether or not the roll-off was smooth my ears could detect that sound of beaming...narrowness if you will, in which the voices made it seem like there were little people singing on the cone.
Even if the roll-off is smooth enough for the KISS principle, I found that the sound was cleaner and more relaxed and refined with at least an inductor in place.
Oh, btw, I generally use 1st-order crossovers on my designs. This is to keep the character of the speakers or the culmination of materials and design relevant to the overall sound of the speaker. i.e., why go carbon fibre without getting a hint of the sound of carbon fibre.
Of course I do this within reason and would work on getting rid of large peaks especially when they are associated with stored energy.
- Status
- Not open for further replies.