Hi Lynn:
I've been following your new thread, and I was wondering if it was still advisable to build a pair of your Mark 6 Ariels? I am on the verge of ordering the VIFA and SCANSPEAK drivers and I wanted to see if there was a high quality crossover 'kit' I could purchase somewhere for the Ariel. I have been thinking about the Ariels for about one year and I've finally found the courage to attempt it.
Should I instead look at an open baffle diy?
Thanks
Mike.
I've been following your new thread, and I was wondering if it was still advisable to build a pair of your Mark 6 Ariels? I am on the verge of ordering the VIFA and SCANSPEAK drivers and I wanted to see if there was a high quality crossover 'kit' I could purchase somewhere for the Ariel. I have been thinking about the Ariels for about one year and I've finally found the courage to attempt it.
Should I instead look at an open baffle diy?
Thanks
Mike.
The referenced CSD is about as bad as it can get. Oddly enough, the frequency-response curves (in a separate chart) are quite a bit better, without the horrifically large nulls and peaks occupying the 3~15 kHz region. The two measurements do not appear to be calculated from the same set of impulse data, from what I can see - or the frequency response is very heavily smoothed to remove the nulls. I usually avoid smoothing for first-arrival time, frequency, and CSD data, just so problems in measurement or the loudspeaker can be more easily seen.
One thing about CSD's is the floor reflection must be absorbed by at least 20~30 dB, otherwise the CSD becomes an unreadable clutter. This takes a mixed pile of pillows, cushions, blankets, and winter coats that is at least 60 cm/2 feet high - and more is better. It's easy enough to build the pile - do a series of repeated impulse measurements, look at the magnitude of the floor reflection around 3~3.5 mSec, and keep adding absorptive material at the location of the floor bounce. They may have failed to do that for this measurement, thus the poor results.
Since the raw impulse data is not shown (it should be, just to get a flavor of the time response) we can't tell if the measurement was valid or not. If it is valid (floor bounce absorbed), then the HF time response is quite poor and resonant. The entire region of the CSD is what I would call unusable and beyond the range of any type of corrective equalization. I hope the author's sake the data is simply nothing more than a measurement corrupted by a floor bounce, or another source of reflection.
A reasonably good CSD is shown below - not the best, but reasonably good. You can read the unsmoothed frequency response of the system by looking at the back of the CSD - see how much smoother it is compared to the other system? Note the absence of sharp, narrow peaks and deep nulls, which are never a good sign in any loudspeaker. I consider this CSD to be a minimum standard for a modern high-fidelity loudspeaker - the new system should exceed this standard, I hope, although it'll probably take a lot of work on the compression driver and horn/waveguide.
P.S. As for building the Ariel Mark 6, it's an pretty old speaker by now. If you are committed to it, why not simplify the task and just build an ME2 with resistive vents? Much easier and very similar sound to the Ariel without the complexity. I've never heard Zaph's speakers, but they certainly measure well in the ways that count, and unlike the Ariel, they use modern drivers that are still in production.
One thing about CSD's is the floor reflection must be absorbed by at least 20~30 dB, otherwise the CSD becomes an unreadable clutter. This takes a mixed pile of pillows, cushions, blankets, and winter coats that is at least 60 cm/2 feet high - and more is better. It's easy enough to build the pile - do a series of repeated impulse measurements, look at the magnitude of the floor reflection around 3~3.5 mSec, and keep adding absorptive material at the location of the floor bounce. They may have failed to do that for this measurement, thus the poor results.
Since the raw impulse data is not shown (it should be, just to get a flavor of the time response) we can't tell if the measurement was valid or not. If it is valid (floor bounce absorbed), then the HF time response is quite poor and resonant. The entire region of the CSD is what I would call unusable and beyond the range of any type of corrective equalization. I hope the author's sake the data is simply nothing more than a measurement corrupted by a floor bounce, or another source of reflection.
A reasonably good CSD is shown below - not the best, but reasonably good. You can read the unsmoothed frequency response of the system by looking at the back of the CSD - see how much smoother it is compared to the other system? Note the absence of sharp, narrow peaks and deep nulls, which are never a good sign in any loudspeaker. I consider this CSD to be a minimum standard for a modern high-fidelity loudspeaker - the new system should exceed this standard, I hope, although it'll probably take a lot of work on the compression driver and horn/waveguide.
P.S. As for building the Ariel Mark 6, it's an pretty old speaker by now. If you are committed to it, why not simplify the task and just build an ME2 with resistive vents? Much easier and very similar sound to the Ariel without the complexity. I've never heard Zaph's speakers, but they certainly measure well in the ways that count, and unlike the Ariel, they use modern drivers that are still in production.
Attachments
J. Bouska is cool. He knows and loves, and recreates earthquakes. Plus his system as a whole, has scientific methodology and clear personal replay preference well met goals. He clearly likes it powerful and electric and shock wave worthy. Goa Trans? Electro? No problem, just hit play.
*I have experienced a very large isobaric sub woofer creation made by a loudspeaker designer friend, (he is much of a multi scientist and an extraordinary pianist) in his home like lab environment, driven with kW amps. Verified it could do a very low distortion 15 Hz 120 dB in there. And I must stress the point that the various recording and structural noises of the building were a very often distraction during replay.
He did it as an experiment just to see what music replay really needed down low, and what not. He listens only to classical and he was very positive that for such repertoire 'it is unnecessary and noisy'.
I guess it takes a specially structured, maybe floating and isolated big room to enjoy such quake grade replay, if into bass head music really.
*I have experienced a very large isobaric sub woofer creation made by a loudspeaker designer friend, (he is much of a multi scientist and an extraordinary pianist) in his home like lab environment, driven with kW amps. Verified it could do a very low distortion 15 Hz 120 dB in there. And I must stress the point that the various recording and structural noises of the building were a very often distraction during replay.
He did it as an experiment just to see what music replay really needed down low, and what not. He listens only to classical and he was very positive that for such repertoire 'it is unnecessary and noisy'.
I guess it takes a specially structured, maybe floating and isolated big room to enjoy such quake grade replay, if into bass head music really.
Hi Lynn and all, Let me start by express my sincere appreciation for a really interresting thread! It has kept me occupied for quite some time.
I am currently restoring a pair of Sonab OA-12 (a pair of theese) and it is kind of funny to read this posting right now.
The OA12:s were designed by Stig Carlsson in the seventies according to his so called "ortho-acoustical principles". One of the cornerstornes in this principle is that the arriving reflections should match the direct sound. Stig Carlsson was head of the electroacoustical laboratory at Royal Acadamy of Tehnology in Sweden and he designed quite good sounding speakers, (quite unconventional too I might add).
His speakers are commonly known as "Carlsson speakers" over here. (Even though most of them was sold via sonab). The OA12's was my first set of HiFi speakers and they had a really nice quality; they didnt sound like you were listening to speakers, rather you listened to music!
I actually didn't know at the time just how good they sounded, It just came to me last year when I discovered Open Baffles that "WoW!, its this kind of transparency again! I've really really missed that!" So at the moment I'm experimenting with open baffles (That is the reason I found this thread) and also renovating my old pair of Carlssons which hasn't played in more than 15 years.
Anyhow, I do think that degradation in the 3-30 ms time window _is_ a bigger issue than commonly known, and I hope you'll appreciate some of the ideas outlined in this paper: http://www.carlssonplanet.com/downloads/index.php?act=view&id=9
Again, thank you everyone for contributing to a gem-sprinkled thread! And to Lynn, I wish you a full and speedy and recovery!
Regards,
Jon Björkebäck
Lynn Olson said:
I am currently restoring a pair of Sonab OA-12 (a pair of theese) and it is kind of funny to read this posting right now.
The OA12:s were designed by Stig Carlsson in the seventies according to his so called "ortho-acoustical principles". One of the cornerstornes in this principle is that the arriving reflections should match the direct sound. Stig Carlsson was head of the electroacoustical laboratory at Royal Acadamy of Tehnology in Sweden and he designed quite good sounding speakers, (quite unconventional too I might add).
His speakers are commonly known as "Carlsson speakers" over here. (Even though most of them was sold via sonab). The OA12's was my first set of HiFi speakers and they had a really nice quality; they didnt sound like you were listening to speakers, rather you listened to music!
I actually didn't know at the time just how good they sounded, It just came to me last year when I discovered Open Baffles that "WoW!, its this kind of transparency again! I've really really missed that!" So at the moment I'm experimenting with open baffles (That is the reason I found this thread) and also renovating my old pair of Carlssons which hasn't played in more than 15 years.
Anyhow, I do think that degradation in the 3-30 ms time window _is_ a bigger issue than commonly known, and I hope you'll appreciate some of the ideas outlined in this paper: http://www.carlssonplanet.com/downloads/index.php?act=view&id=9
Again, thank you everyone for contributing to a gem-sprinkled thread! And to Lynn, I wish you a full and speedy and recovery!
Regards,
Jon Björkebäck
Madison Warrior
Just a little heads up on Warrior drivers. During 2 weeks of phoning Warrior a few times and Shred many times no one ever answered the phones at either place. I left messages, finally Shred returned my call and I ordered 10 12" warriors. That was Dec 12. I recieved an order confirmation and my CC was billed. I expected to have drivers within 2 weeks. New years came and went and still no drivers. I'm back on the phone and leaving messages again. No response. My CC statment shows the charges were reversed 7 days after the order was confirmed. To this day no one from Shred has contacted me with any explanation. I lost 30 bucks on the exchange rate, waited weeks for drivers that were never shipped and Shred didn't have the decency to phone or email and let me know what they did! I can't see these guys lasting!
I did what I should have done to begin with, I ordered 4 Selenium 15PW3-SLF 15's from Parts Express. I running them in a bi-amped system on ~24 x 32 U baffles with 11" wings. They have completely exceeded my expectations. They are extremely agile, beautiful tone and clarity with tremendous power and delicacy
Lynn Olson said:
Just a little heads up on Warrior drivers. During 2 weeks of phoning Warrior a few times and Shred many times no one ever answered the phones at either place. I left messages, finally Shred returned my call and I ordered 10 12" warriors. That was Dec 12. I recieved an order confirmation and my CC was billed. I expected to have drivers within 2 weeks. New years came and went and still no drivers. I'm back on the phone and leaving messages again. No response. My CC statment shows the charges were reversed 7 days after the order was confirmed. To this day no one from Shred has contacted me with any explanation. I lost 30 bucks on the exchange rate, waited weeks for drivers that were never shipped and Shred didn't have the decency to phone or email and let me know what they did! I can't see these guys lasting!
I did what I should have done to begin with, I ordered 4 Selenium 15PW3-SLF 15's from Parts Express. I running them in a bi-amped system on ~24 x 32 U baffles with 11" wings. They have completely exceeded my expectations. They are extremely agile, beautiful tone and clarity with tremendous power and delicacy
Lynn's blog post got me thinking 
about time alignment. I'm inclined to believe that arrival time and time coherence are extremely important for realistic sound reproduction. However noble that goal may be it seems equally elusive. We are dealing with tallish speakers that place drives well above and below the listeners ears. Time aligning them horizontally does not align them to the listener. The baffle would have to be multi angled or curved to align the drivers to each other and a spot exactly on axis for all drivers at a fixed and equal distance and height (where my ear is) from each driver.
Even if I find the acoustic center of a driver I expect it changes with frequency.
Please note that I am not stating any of the above as fact, mearly what makes sense to me at this time.
Regards
about time alignment. I'm inclined to believe that arrival time and time coherence are extremely important for realistic sound reproduction. However noble that goal may be it seems equally elusive. We are dealing with tallish speakers that place drives well above and below the listeners ears. Time aligning them horizontally does not align them to the listener. The baffle would have to be multi angled or curved to align the drivers to each other and a spot exactly on axis for all drivers at a fixed and equal distance and height (where my ear is) from each driver. Even if I find the acoustic center of a driver I expect it changes with frequency.
Please note that I am not stating any of the above as fact, mearly what makes sense to me at this time.
Regards
Re: Madison Warrior
Sorry to hear about getting screwed - I bought my 10's a while ago from 'Steel Sound' - not sure if it's the same people.
Those Seliniums must sound great!
riff.ca said:
Just a little heads up on Warrior drivers. During 2 weeks of phoning Warrior a few times and Shred many times no one ever answered the phones at either place. I left messages, finally Shred returned my call and I ordered 10 12" warriors. That was Dec 12. I recieved an order confirmation and my CC was billed. I expected to have drivers within 2 weeks. New years came and went and still no drivers. I'm back on the phone and leaving messages again. No response. My CC statment shows the charges were reversed 7 days after the order was confirmed. To this day no one from Shred has contacted me with any explanation. I lost 30 bucks on the exchange rate, waited weeks for drivers that were never shipped and Shred didn't have the decency to phone or email and let me know what they did! I can't see these guys lasting!
I did what I should have done to begin with, I ordered 4 Selenium 15PW3-SLF 15's from Parts Express. I running them in a bi-amped system on ~24 x 32 U baffles with 11" wings. They have completely exceeded my expectations. They are extremely agile, beautiful tone and clarity with tremendous power and delicacy
Sorry to hear about getting screwed - I bought my 10's a while ago from 'Steel Sound' - not sure if it's the same people.
Those Seliniums must sound great!
Re: Re: Re: Re: Madison Warrior
Its a simple 3 way sys; Bass driver is sealed 110 Lts crossed a 400hz 1st order to an FE164 and 2nd order at 10k to a Chinese titanium 1" "bullet". All passive xo.
The system sounds very good, but of course nothing to do of what I read of your outstanding OB systems, congrats for those.
Magnetar said:
Its a simple 3 way sys; Bass driver is sealed 110 Lts crossed a 400hz 1st order to an FE164 and 2nd order at 10k to a Chinese titanium 1" "bullet". All passive xo.
The system sounds very good, but of course nothing to do of what I read of your outstanding OB systems, congrats for those.
Lynn Olson said:
Lynn,
here's CSD of my current mains - VAF i66. They do sound as clean as this graph suggests (they use twin Seas magnesum Excels as mids), and in this respect they indeed rival ATC's mighty SCM75S.
What this graph doesn't show is how 'dry' they sound compared to ATC. Don't know what it is, but ATC simply sounds more like real music. VAF in direct comparison, well, sounds like (admitedly excellent) HiFi. I really would like to get to the bottom of the technical explanation of what really causes this, but so far I have failed.
Bratislav
Attachments
To Bratislav: That is an excellent CSD, no question about it. But you're right, it doesn't tell you anything about tone quality - speakers with pretty poor CSD's can have outstanding tone quality (the Lowther and various 1950's classics come to mind here), and speakers with amazingly good CSD's can sound dry and analytic, with little musical flavor and pale, washed-out tonality.
You can also have speakers that are bad in every way, bad CSD's, terrible FR, no tonality, flattened dynamics, no imaging, grossly colored, frankly just crummy all around - and some of these at truly astounding prices, in the $10,000 to $100,000/pair range. And there are very rare speakers that have good CSD's and superb tonality - the finest studio monitors and the outer limits of the DIY world comes to mind here. Almost no commercial high-end speaker would fall in this group, at any price.
A good CSD is associated with absence of gross coloration, natural-sounding vocals and violins, and very realistic spatial qualities - I feel rapid settling in the time domain is more important than the polar pattern for spatial realism. It says very little whether the tone quality is vivid, juicy, exciting, or pale, washed-out, and boring to listen to.
The vividness of tone is more elusive in a technical sense - the most vivid speakers tend to be quite efficient, have sophisticated magnetic design (Alnico helps), lighter cones, higher BL products, fairly simple crossovers - and, well, a certain something, a consistent level of sophisticated and thorough engineering for the drivers. Of the speakers I've heard for myself, large-format Altec, Alnico Lowthers, Alnico Tone Tubby, the field-coil Cogent, AudioKinesis at the 2007 RMAF, and the field-coil and Alnico Feastrex all come to mind as having a thrilling, exciting sound that really brings the music to life - and a sound that is immediately recognizable and brings pleasant memories to mind.
The difficult part is bringing the two worlds together - I don't think they are inherently opposed, they just come from two different audio-cultures, and both have worthwhile contributions to bring to the table.
To Riff.ca: Time alignment and rapid decay (good CSD) are two different things. Time alignment is typically thought of as the ability to reproduce recognizable square waves, or in other words, a single recognizable impulse. You can have a system with quite distorted time response - say, with 4th-order crossovers, which act as 2nd-order allpass filters, and quite warped-looking square waves - that has excellent CSD's and rapid settling characteristics. This is quite common in high-quality studio monitors, as well as Siegfried Linkwitz's speakers.
There's still controversy over how much time distortion is audible - I suspect there is a lot personal variation on this one, some people are very sensitive, others barely hear it at all. Speaking for myself, I hear a slightly audible variation in timbre when the all-pass switch is flipped - others hear a dramatic good/bad change in the sound.
That being said, almost all Golden-Age horn systems and many contemporary horn systems have atrocious time distortion, because it looks a lot prettier to align the horn mouths all together. Bass horns and supertweeter horns are the worst offenders here. Although it is ugly looking, what should be closely aligned are the diaphragms, not the horn mouths. Of course, that looks weird, especially if the tweeter ends up in the next room so it can be as far away as the basshorn diaphragm.
A time error of a fraction of wavelength is a lot smaller concern than many wavelengths, which is where a big multihorn system gets in trouble. I've tried to align a horn crossover with the tweeter misaligned by several wavelengths, and it's a big mess, with lots of ripples and terrible off-axis response. All the hassles disappeared when the diaphragms were placed in acoustically close alignment with instrumentation. (See my simple experiments with the Klipsch Chorus here.) I should add that time-sensitive (no RTA's) instrumentation is required for any kind of time-alignment - you can't do it just by looking, sorry.
As for the physical construction of the new system, that's pretty straightforward. The ribbon supertweeter will be on top, supported by a vertical pole, and the compression driver+horn/waveguide will be lower on the same vertical pole, with the centerline between 36 and 42 inches high - the center of the horn/waveguide will be on the visual horizon-line. Both drivers will be in-line vertically, and will have fine adjustments for front-to-back alignment. These adjustments will be done by measurement (MLSSA) at the listening distance of 3~5 meters.
The dipole 12" midbass driver(s) will be directly below - this will be 18Sound 12NDA520's, Altec/GPA 414's, or 12" Tone Tubbies - to be determined by audition and measurement. There will be two side wings, each with a pair of 12 or 15-inch high-Q bass-fill drivers.
1 VHF ribbon, 1 mid compression driver, most likely 2 midbass drivers, and 4 bass-fill drivers, on either side of the vertical midbass pair. Total radiating area for the direct-radiators will be 500 square inches or more, depending on the choice of 12 or 15-inch drivers for the bass-fill application. (The bass-fill drivers will have their own amplification and equalization.)
A minimalist version of the same system (for smaller rooms) would retain the VHF and mid drivers, but use a single 12" midbass driver directly below the mid drivers, and two side-by-side 15" bass-fill drivers at the bottom. Basically, mix-n-match for the application.
You can also have speakers that are bad in every way, bad CSD's, terrible FR, no tonality, flattened dynamics, no imaging, grossly colored, frankly just crummy all around - and some of these at truly astounding prices, in the $10,000 to $100,000/pair range. And there are very rare speakers that have good CSD's and superb tonality - the finest studio monitors and the outer limits of the DIY world comes to mind here. Almost no commercial high-end speaker would fall in this group, at any price.
A good CSD is associated with absence of gross coloration, natural-sounding vocals and violins, and very realistic spatial qualities - I feel rapid settling in the time domain is more important than the polar pattern for spatial realism. It says very little whether the tone quality is vivid, juicy, exciting, or pale, washed-out, and boring to listen to.
The vividness of tone is more elusive in a technical sense - the most vivid speakers tend to be quite efficient, have sophisticated magnetic design (Alnico helps), lighter cones, higher BL products, fairly simple crossovers - and, well, a certain something, a consistent level of sophisticated and thorough engineering for the drivers. Of the speakers I've heard for myself, large-format Altec, Alnico Lowthers, Alnico Tone Tubby, the field-coil Cogent, AudioKinesis at the 2007 RMAF, and the field-coil and Alnico Feastrex all come to mind as having a thrilling, exciting sound that really brings the music to life - and a sound that is immediately recognizable and brings pleasant memories to mind.
The difficult part is bringing the two worlds together - I don't think they are inherently opposed, they just come from two different audio-cultures, and both have worthwhile contributions to bring to the table.
To Riff.ca: Time alignment and rapid decay (good CSD) are two different things. Time alignment is typically thought of as the ability to reproduce recognizable square waves, or in other words, a single recognizable impulse. You can have a system with quite distorted time response - say, with 4th-order crossovers, which act as 2nd-order allpass filters, and quite warped-looking square waves - that has excellent CSD's and rapid settling characteristics. This is quite common in high-quality studio monitors, as well as Siegfried Linkwitz's speakers.
There's still controversy over how much time distortion is audible - I suspect there is a lot personal variation on this one, some people are very sensitive, others barely hear it at all. Speaking for myself, I hear a slightly audible variation in timbre when the all-pass switch is flipped - others hear a dramatic good/bad change in the sound.
That being said, almost all Golden-Age horn systems and many contemporary horn systems have atrocious time distortion, because it looks a lot prettier to align the horn mouths all together. Bass horns and supertweeter horns are the worst offenders here. Although it is ugly looking, what should be closely aligned are the diaphragms, not the horn mouths. Of course, that looks weird, especially if the tweeter ends up in the next room so it can be as far away as the basshorn diaphragm.
A time error of a fraction of wavelength is a lot smaller concern than many wavelengths, which is where a big multihorn system gets in trouble. I've tried to align a horn crossover with the tweeter misaligned by several wavelengths, and it's a big mess, with lots of ripples and terrible off-axis response. All the hassles disappeared when the diaphragms were placed in acoustically close alignment with instrumentation. (See my simple experiments with the Klipsch Chorus here.) I should add that time-sensitive (no RTA's) instrumentation is required for any kind of time-alignment - you can't do it just by looking, sorry.
As for the physical construction of the new system, that's pretty straightforward. The ribbon supertweeter will be on top, supported by a vertical pole, and the compression driver+horn/waveguide will be lower on the same vertical pole, with the centerline between 36 and 42 inches high - the center of the horn/waveguide will be on the visual horizon-line. Both drivers will be in-line vertically, and will have fine adjustments for front-to-back alignment. These adjustments will be done by measurement (MLSSA) at the listening distance of 3~5 meters.
The dipole 12" midbass driver(s) will be directly below - this will be 18Sound 12NDA520's, Altec/GPA 414's, or 12" Tone Tubbies - to be determined by audition and measurement. There will be two side wings, each with a pair of 12 or 15-inch high-Q bass-fill drivers.
1 VHF ribbon, 1 mid compression driver, most likely 2 midbass drivers, and 4 bass-fill drivers, on either side of the vertical midbass pair. Total radiating area for the direct-radiators will be 500 square inches or more, depending on the choice of 12 or 15-inch drivers for the bass-fill application. (The bass-fill drivers will have their own amplification and equalization.)
A minimalist version of the same system (for smaller rooms) would retain the VHF and mid drivers, but use a single 12" midbass driver directly below the mid drivers, and two side-by-side 15" bass-fill drivers at the bottom. Basically, mix-n-match for the application.
soongsc said:
But in this case I have two speakers with excellent CSD, and I prefer one whose drivers are not stiff (ATC's mid is soft and sticky, while Excels' magnesium drivers are as stiff as it gets).
And I'm definitely not from the SET camp, I do care about realistic sound, not harmonics enriched, "pink colored glasses" sound.
Live sound is always my reference, and I try to hear it as often as I can (at least once a month, usually much more often).
What gives ?
