Since I already have the superb RAAL tweeters, I intend to use them in the region where horns start to become directive (throat circumference approaches a wavelength) and compression drivers rely on bandwidth-extending tricks like broad regions of resonance (which also greatly increases distortion). The extended HF is a difficult region for horn/compression drivers, unless it is a dedicated supertweeter - another breed of cat entirely.
By relieving the horn system above 5~7 kHz, a larger throat diameter is acceptable. This allows either a small cone driver with a 3.5" throat (similar to what Magnetar is using), or a large-format (1.4, 1.5, or 2") compression driver. Both have substantially improved power-handling, and more importantly, lower IM distortion compared to a 1" small-format compression driver, particularly in the critical 1~3 kHz region, where the ear reaches peak sensitivity for distortion and time-domain errors.
The 1~3 kHz region is the most critical part of the spectrum, and design choices in this region make the difference between a loudspeaker that is high fidelity and one that is not. Unfortunately for direct-radiators, it is at the top of the working range, and the region where cone breakup is starting to make its presence known. Similarly, for small-format compression drivers, it is at the bottom of the working range, and a region where distortion is rising rapidly below 1 kHz. Trust me, when a driver is working at the end of its frequency range, it is easily audible - as a sense of "stress", unease, and a lurking potential for sudden harshness at unexpected times. Systems that don't do this are systems we call "relaxed" and confident-sounding.
This is the primary reason I am interested in a large-format horn with an additional HF driver. The 1~3 kHz region needs to be served with the highest-quality technology available - the lowest IM distortion, and the cleanest time-domain decay characteristics (more than 30 dB down in less than 1 mSec is desirable).
There are not many choices here - and all of them will have a clear and distinct "personality". I do not know how the 1.4" compression drivers will compare to prosound cone drivers - that will take extended listening to make an assessment.
By relieving the horn system above 5~7 kHz, a larger throat diameter is acceptable. This allows either a small cone driver with a 3.5" throat (similar to what Magnetar is using), or a large-format (1.4, 1.5, or 2") compression driver. Both have substantially improved power-handling, and more importantly, lower IM distortion compared to a 1" small-format compression driver, particularly in the critical 1~3 kHz region, where the ear reaches peak sensitivity for distortion and time-domain errors.
The 1~3 kHz region is the most critical part of the spectrum, and design choices in this region make the difference between a loudspeaker that is high fidelity and one that is not. Unfortunately for direct-radiators, it is at the top of the working range, and the region where cone breakup is starting to make its presence known. Similarly, for small-format compression drivers, it is at the bottom of the working range, and a region where distortion is rising rapidly below 1 kHz. Trust me, when a driver is working at the end of its frequency range, it is easily audible - as a sense of "stress", unease, and a lurking potential for sudden harshness at unexpected times. Systems that don't do this are systems we call "relaxed" and confident-sounding.
This is the primary reason I am interested in a large-format horn with an additional HF driver. The 1~3 kHz region needs to be served with the highest-quality technology available - the lowest IM distortion, and the cleanest time-domain decay characteristics (more than 30 dB down in less than 1 mSec is desirable).
There are not many choices here - and all of them will have a clear and distinct "personality". I do not know how the 1.4" compression drivers will compare to prosound cone drivers - that will take extended listening to make an assessment.
Re: Re: Re: More AH-550 Information
When it comes to horns it would be a GROSS error to "forget off axis".
As to the response at 1m or 2m - no doubt what you would expect, a "falling" high freq. response as distance increases. Still important though! (..something Lynn will have to sort out when crossing over to the RAAL.)
john k... said:
When it comes to horns it would be a GROSS error to "forget off axis".
As to the response at 1m or 2m - no doubt what you would expect, a "falling" high freq. response as distance increases. Still important though!
(..something Lynn will have to sort out when crossing over to the RAAL.)
Neither the RAAL nor the mid-horn will be following the inverse-square distance law above 7 kHz, so output at a listening distance of 3~5 meters will be substantially higher than an equivalent dome tweeter radiating into a hemisphere. With a 1.4" throat, the horn will be at the top of its resistive-radiation range (just starting to beam), and the RAAL will have very wide horizontal (approaching 180 degrees) and fairly narrow vertical dispersion. Both result in additional mid and HF output at greater listening distances.
The narrow vertical dispersion of the RAAL will require some kind of system to point it up and down - maybe nothing more than a set of wooden shims at the rear of the magnet assembly.
I will be using autoformers with 1 dB taps (in the same style as Klipsch, if not the actual Klipsch parts) for trimming mid/HF levels to listening distance. As ScottG mentions, the mid horn may require mild shelf equalization - this circuit would precede the autoformers and be part of the mid/high crossover.
The crossover/autoformer topology is: LP/HP/Shelf filter -> 8 ohm shunt resistor -> high Z primary -> low Z secondary -> compression driver. The autoformer multiplies the Z curve of the compression driver many times, and the 8 ohm parallel shunt resistor flattens out the variations to a value only slightly above 8 ohms. The autoformer also provides a low source resistance to the compression driver, effectively making the source Z of the crossover disappear from the view of the compression driver.
I'll be taking a brief break from DIYAudio and writing an article for the Clarisonus blog entitled "A Different View of Dispersion". Rather than tediously explaining my view over and over again in this thread, I'll write the thing once and just point to John Atwood's blog, where it'll stay. This is a contrarian view largely at odds with the industry-prevailing Floyd Toole approach.
The narrow vertical dispersion of the RAAL will require some kind of system to point it up and down - maybe nothing more than a set of wooden shims at the rear of the magnet assembly.
I will be using autoformers with 1 dB taps (in the same style as Klipsch, if not the actual Klipsch parts) for trimming mid/HF levels to listening distance. As ScottG mentions, the mid horn may require mild shelf equalization - this circuit would precede the autoformers and be part of the mid/high crossover.
The crossover/autoformer topology is: LP/HP/Shelf filter -> 8 ohm shunt resistor -> high Z primary -> low Z secondary -> compression driver. The autoformer multiplies the Z curve of the compression driver many times, and the 8 ohm parallel shunt resistor flattens out the variations to a value only slightly above 8 ohms. The autoformer also provides a low source resistance to the compression driver, effectively making the source Z of the crossover disappear from the view of the compression driver.
I'll be taking a brief break from DIYAudio and writing an article for the Clarisonus blog entitled "A Different View of Dispersion". Rather than tediously explaining my view over and over again in this thread, I'll write the thing once and just point to John Atwood's blog, where it'll stay. This is a contrarian view largely at odds with the industry-prevailing Floyd Toole approach.
Re: Re: Re: Re: More AH-550 Information
I'm not saying forget about it. Just making the point that a measuremenet at the mouth isn't eaxctly what I would expect at 1M, or further out, and I don't mean just roll off of high frequencies.
ScottG said:
When it comes to horns it would be a GROSS error to "forget off axis".
I'm not saying forget about it. Just making the point that a measuremenet at the mouth isn't eaxctly what I would expect at 1M, or further out, and I don't mean just roll off of high frequencies.
Re: Re: Re: More AH-550 Information
5
No. Forget on and off axis, just EnaBL the darn thing(s).
Same for the RAAL. EnaBL it.
If such stellar cone designs as Fostex, Lowther, Radioshack, etc. can possibly be improved upon by added mass stiffening/damping, certainly the ribbon here can also be taken to the next level (up of course). Supposedly even horns and baffles (as to be used here) are not immune to its benefits?
I'm now thinking of placing some EnaBling directly upon my ears, to help me actually hear better?
The off axis, DI, etc of this design is definitely not going to follow the Toole, JBL, etc. research. Some serious wavelength spacing between AC's too.
cheers,
AJ
5
john k... said:
No. Forget on and off axis, just EnaBL the darn thing(s).
Same for the RAAL. EnaBL it.
If such stellar cone designs as Fostex, Lowther, Radioshack, etc. can possibly be improved upon by added mass stiffening/damping, certainly the ribbon here can also be taken to the next level (up of course). Supposedly even horns and baffles (as to be used here) are not immune to its benefits?
I'm now thinking of placing some EnaBling directly upon my ears, to help me actually hear better?
The off axis, DI, etc of this design is definitely not going to follow the Toole, JBL, etc. research. Some serious wavelength spacing between AC's too.
cheers,
AJ
Hello Lynn
"I am in agreement with Newell and Dr. Holland on essentially all of the points. I'm primarily interested in smooth FR, time/phase response, and directivity (most importantly, without narrow spikes only a few degrees wide). I'm less interested in a specified degree of directivity than I am in freedom from narrow artifacts in frequency, time, or spatial domains - thus the interest in drivers that are free from breakup in the working band, horn/waveguides that are free of internal obstructions, and phase-plugs that emit coherent wavefronts. Smooth driver response without a lot of wacko equalization is a major advantage in my book - and I'm aware this isn't a fashionable viewpoint in the pro and audiophile worlds these days."
If something is fashionable or not is really not important. The reason the SR community at large embraces Constant Directivity and CD type wave guides is because they make so much sense in that application.
I think you should be looking at what's happening off axis and particularly what the directivity is around your proposed crossover points. All the phase, frequency and time anomalies you are trying to avoid potentially will be present is a 3 way using midrange and tweeter components that have narrowing directivity patterns on axis. Just the diameter of the horn and the voice coils offsets will almost guarantee comb filtering through the proposed crossover range. All you need to do is look at the various system measurements from the classic JBL 3 and 4 way monitors.
If you look at the 2420/2307/2312 compression driver horn combo the directivity of the chosen horn has to be close to them to flatten the on axis response of the 2420. The directivity of the 2405 is similar to what the ribbons are far a directivity is concerned with narrow in the vertical and wide horizontal axis. Both the horn and 2405's directivity were very limited through the crossover range and above.
If you look at the monitors the 2405 and 2307 are side by side. Looking at the directivity it easy to see why. These systems sound the most balanced when your ears are on axis with these drivers. Putting them side by side cure this does not effect the comb filtering through the crossover range.
The best way to avoid this in simply not add a tweeter. It's name your poison though as you don't want to run a 1" driver to low. If you look at the 4 ways, which are arguably the best sounding a 10" midrange covers the 300hz-1.2K range. It's all about compromises.
If you have not seen these take a minute to read them. You will enjoy them just for the read
http://audioheritage.org/vbulletin/showthread.php?t=7852
http://audioheritage.org/vbulletin/showthread.php?t=6687
http://audioheritage.org/vbulletin/showthread.php?t=4410
http://audioheritage.org/vbulletin/showthread.php?t=4408
Attached is a system plot for the 4333.
Just food for thought.
Rob
"I am in agreement with Newell and Dr. Holland on essentially all of the points. I'm primarily interested in smooth FR, time/phase response, and directivity (most importantly, without narrow spikes only a few degrees wide). I'm less interested in a specified degree of directivity than I am in freedom from narrow artifacts in frequency, time, or spatial domains - thus the interest in drivers that are free from breakup in the working band, horn/waveguides that are free of internal obstructions, and phase-plugs that emit coherent wavefronts. Smooth driver response without a lot of wacko equalization is a major advantage in my book - and I'm aware this isn't a fashionable viewpoint in the pro and audiophile worlds these days."
If something is fashionable or not is really not important. The reason the SR community at large embraces Constant Directivity and CD type wave guides is because they make so much sense in that application.
I think you should be looking at what's happening off axis and particularly what the directivity is around your proposed crossover points. All the phase, frequency and time anomalies you are trying to avoid potentially will be present is a 3 way using midrange and tweeter components that have narrowing directivity patterns on axis. Just the diameter of the horn and the voice coils offsets will almost guarantee comb filtering through the proposed crossover range. All you need to do is look at the various system measurements from the classic JBL 3 and 4 way monitors.
If you look at the 2420/2307/2312 compression driver horn combo the directivity of the chosen horn has to be close to them to flatten the on axis response of the 2420. The directivity of the 2405 is similar to what the ribbons are far a directivity is concerned with narrow in the vertical and wide horizontal axis. Both the horn and 2405's directivity were very limited through the crossover range and above.
If you look at the monitors the 2405 and 2307 are side by side. Looking at the directivity it easy to see why. These systems sound the most balanced when your ears are on axis with these drivers. Putting them side by side cure this does not effect the comb filtering through the crossover range.
The best way to avoid this in simply not add a tweeter. It's name your poison though as you don't want to run a 1" driver to low. If you look at the 4 ways, which are arguably the best sounding a 10" midrange covers the 300hz-1.2K range. It's all about compromises.
If you have not seen these take a minute to read them. You will enjoy them just for the read
http://audioheritage.org/vbulletin/showthread.php?t=7852
http://audioheritage.org/vbulletin/showthread.php?t=6687
http://audioheritage.org/vbulletin/showthread.php?t=4410
http://audioheritage.org/vbulletin/showthread.php?t=4408
Attached is a system plot for the 4333.
Just food for thought.
Rob
Attachments
Since the target application is not a soffit-mounted studio monitor, there is no requirement for the midbass driver mounting flange, mid-horn mouth, and ribbon tweeter flange to be on a common plane. Although physically convenient for a conventional box speaker, there's no reason to do this when the mid and HF drivers are mounted in free space (instead of a front baffle). The preliminary system will have the acoustic centers of the drivers at a common distance from the listening location - to be determined by measurement.
In my experience, nulls in response at frequencies of 7 kHz or above are surprisingly difficult to hear (unlike 3 kHz or lower). This is what makes supertweeters practical; although the supertweeter and the mid or tweeter below it are spaced several wavelengths apart at the crossover frequency, audibility of the null as you move up and down is surprisingly small - even with pink-noise stimulus, the most sensitive test material. With music, audibility is even lower.
That luxury does not exist at the lower-frequency crossover in the 640~800 Hz range, where lobing is quite audible. Thus the provision of a single 15" driver to cover the 50~800 Hz range, with a mid-horn mounted directly above it. The other 15" drivers are only active at much longer wavelengths.
The mouth of the mid-horn is probably not going to be on the same plane as the midbass baffle; it will probably be several inches in front or in back of the midbass baffle - depending on the alignment of the acoustic centers of the midbass driver and the mid-horn (which is affected in the turn by the electrical delay of the lowpass crossover for the midbass driver). To be determined by measurement, although I expect the acoustic centers to be fairly close to the voice-coils.
My goals, as mentioned earlier, are not compliant with Toole (et al) directivity criteria, and are not suitable for studio-monitor or THX applications. I am aiming for the first-arrival sound to have flat frequency response, rapid time decay, and low IM distortion across a horizontal arc aimed at the listening area. The overall power response into a sphere, or 90-degree cone, although important, is less important than the first-arrival performance.
I am grateful every day I don't have to report to Harmon International, a corporate marketing department, or apply for THX certification. Frankly, if I had to report to any of these organizations, I'd get out of hifi altogether, and do something completely different.
In my experience, nulls in response at frequencies of 7 kHz or above are surprisingly difficult to hear (unlike 3 kHz or lower). This is what makes supertweeters practical; although the supertweeter and the mid or tweeter below it are spaced several wavelengths apart at the crossover frequency, audibility of the null as you move up and down is surprisingly small - even with pink-noise stimulus, the most sensitive test material. With music, audibility is even lower.
That luxury does not exist at the lower-frequency crossover in the 640~800 Hz range, where lobing is quite audible. Thus the provision of a single 15" driver to cover the 50~800 Hz range, with a mid-horn mounted directly above it. The other 15" drivers are only active at much longer wavelengths.
The mouth of the mid-horn is probably not going to be on the same plane as the midbass baffle; it will probably be several inches in front or in back of the midbass baffle - depending on the alignment of the acoustic centers of the midbass driver and the mid-horn (which is affected in the turn by the electrical delay of the lowpass crossover for the midbass driver). To be determined by measurement, although I expect the acoustic centers to be fairly close to the voice-coils.
My goals, as mentioned earlier, are not compliant with Toole (et al) directivity criteria, and are not suitable for studio-monitor or THX applications. I am aiming for the first-arrival sound to have flat frequency response, rapid time decay, and low IM distortion across a horizontal arc aimed at the listening area. The overall power response into a sphere, or 90-degree cone, although important, is less important than the first-arrival performance.
I am grateful every day I don't have to report to Harmon International, a corporate marketing department, or apply for THX certification. Frankly, if I had to report to any of these organizations, I'd get out of hifi altogether, and do something completely different.
As mentioned earlier, I don't know if the Great Plains Audio 416 and 515 Alnico drivers will available any time soon. I'll be buying one or the other when they become available. I have 18Sound 15NMB420's on their way, and these are akin to Altec 416's and 515's - although it's a safe guess they probably sound different.
The 18Sound 15MB700 appears a bit more akin to the 515, with a noticeably rising response, unlike the 15NMB420. You'd think they would almost be identical twins, with 73-gram cones, but the cones are evidently different based on the photos and measurements.
Regarding ribbons as supertweeters, others have reported good results with compression drivers. If they can work successfully with a driver as awkward as a Lowther, a compression driver should be more straightforward, especially if it is aligned in the time domain. (If the drivers are misaligned in the time domain by a millisecond or more, as is typical in conventional enclosure multihorn systems, then crossover design gets a lot more difficult.)
The 18Sound 15MB700 appears a bit more akin to the 515, with a noticeably rising response, unlike the 15NMB420. You'd think they would almost be identical twins, with 73-gram cones, but the cones are evidently different based on the photos and measurements.
Regarding ribbons as supertweeters, others have reported good results with compression drivers. If they can work successfully with a driver as awkward as a Lowther, a compression driver should be more straightforward, especially if it is aligned in the time domain. (If the drivers are misaligned in the time domain by a millisecond or more, as is typical in conventional enclosure multihorn systems, then crossover design gets a lot more difficult.)
Sobering Article in Rolling Stone
The Death of High Fidelity.
Quite a counterpoint to the CES madness this week, isn't it? More on the topic here.
After all the fanfare and industry buzz about the wonderful convenience and versatility of software-based Digital Audio Workstations, this is what the software is actually used for - extreme dynamic compression that was technically impossible back in the all-analog days.
The Death of High Fidelity.
Quite a counterpoint to the CES madness this week, isn't it? More on the topic here.
After all the fanfare and industry buzz about the wonderful convenience and versatility of software-based Digital Audio Workstations, this is what the software is actually used for - extreme dynamic compression that was technically impossible back in the all-analog days.
Lynn Olson said:
Amen to that!
The Revel line..
Makes me think of the Edsel for the price of a Rolls.
It would seem that all their engineers and test groups have some how managed to engineer complete mediocrity for the masses at a price only fools would pay. And yet - they consistently receive really good "reviews" in several magazines.. and unlike conspiracy buffs, I DON'T think the magazine is being directly subverted for the sake of adv. cash, rather I think there are some amazingly bad reviewers out there with the occasional editors that will "let it slide" for the sake of adv. cash. (ok, only a slight difference there.
) Sadly, there are lots of highly acclaimed products that I think similarly of (dcs for one) - but typically without the VAST R&D backing of Harmon.
Hmm, 14 years ago you wouldn't need to use all your fingers to count the number of loudspeakers that were priced over 10k. Today a recent "paper-based" publication declared the latest "reference" from Revel to be something of a BARGAIN (at a paltry 22.5k ..if I remember correctly).
..well end of rant.
Great thing about all this is the "push" to DIY and the vast array of products available (..even if most of them are often priced in the nuttysphere).
Re: Re: Gradient Helsinki 1.5
I highly doubt that. Where are you getting this information from? Do you have a link? I don't see any way that would be an open slot.
It appears to me as I originally described, a corner dipole, with a solid baffle, similar to what SL's example on his site (but a single baffle). Should be usable away from the corner with a corresponding loss of LF bass efficiency.
I have heard many Revels (sold locally). They are on par with any box design I have heard (I have not yet heard the TAD Model-1). Not my cup of tea, but as good or better than the typical monkey coffins, most of which are awful.
cheers,
AJ
ScottG said:
I highly doubt that. Where are you getting this information from? Do you have a link? I don't see any way that would be an open slot.
It appears to me as I originally described, a corner dipole, with a solid baffle, similar to what SL's example on his site (but a single baffle). Should be usable away from the corner with a corresponding loss of LF bass efficiency.
I have heard many Revels (sold locally). They are on par with any box design I have heard (I have not yet heard the TAD Model-1). Not my cup of tea, but as good or better than the typical monkey coffins, most of which are awful.
cheers,
AJ
Re: Re: Re: Gradient Helsinki 1.5
The gradient - I think I remember it from show pictures (the rear of it). It was a while ago.
The Revel's I've heard have linearity, and often extension. Tonally they are sorely lacking and they have a good deal of what I think of "early on-set compression" (i.e. "dull and dynamically constricted"). They lack soundstage depth and width, and performers/instruments are typically "compressed" together. Like most "monkey coffins" they seemingly "push" drum kits forward, sometimes superimposing on the other performers and even occasionally in front of the the other performers. Clean, but uninvolving.
You can get better overall sound from a pair of Vandersteen 2ce's for a LOT less despite the fact that they aren't as linear or have as much extension (..and they have similar problems with tonality and early on-set compression).
AJinFLA said:
The gradient - I think I remember it from show pictures (the rear of it). It was a while ago.
The Revel's I've heard have linearity, and often extension. Tonally they are sorely lacking and they have a good deal of what I think of "early on-set compression" (i.e. "dull and dynamically constricted"). They lack soundstage depth and width, and performers/instruments are typically "compressed" together. Like most "monkey coffins" they seemingly "push" drum kits forward, sometimes superimposing on the other performers and even occasionally in front of the the other performers. Clean, but uninvolving.
You can get better overall sound from a pair of Vandersteen 2ce's for a LOT less despite the fact that they aren't as linear or have as much extension (..and they have similar problems with tonality and early on-set compression).
Hi
Lynn, congratulation to jump 3000 postings in your thread.
Marrying a small diameter driver horn at ~ 800 Hz at considerable SPL AND low IM AND a low order XO will become an interesting exercise .
I am currently figuring out a way to import the results of the "horn response analysis program" HORNRESP from David J McBean into my IM sheet.
This should deliver some nice insights about what HP filter orders to use at minimum for horns and wave guides .
Greetings
Michael
Lynn, congratulation to jump 3000 postings in your thread.
Marrying a small diameter driver horn at ~ 800 Hz at considerable SPL AND low IM AND a low order XO will become an interesting exercise .
I am currently figuring out a way to import the results of the "horn response analysis program" HORNRESP from David J McBean into my IM sheet.
This should deliver some nice insights about what HP filter orders to use at minimum for horns and wave guides .
Greetings
Michael
Re: Sobering Article in Rolling Stone
Great article. Great composers know where to put the silence between the notes. Now the silence has veen eliminated. Explains where so much of the new music is boring.
Also reminds me of a conversation with Roger West where he explained about why tubes sound better than transistors by talking about the waveforms and the way transistors and tubes clip the waveform differently. The type of clipping discussed in the srticle would generate tons of high order harmonic distortion, A topic you have written quite a bit about eh Lynn. The shame is that we could have morwe dynamic cintrasts rather than less.
I also wonder about the death wish the music industry has, they do everything possible to make a buck rather than make good recordings and then complain when people download compressed recordings of the compressed music they create, since the sound doesn't really matter.... or people just don't care since the recording is no better than what they hear on MTV or radio, so why pay for the recording?
Lynn Olson said:
Great article. Great composers know where to put the silence between the notes. Now the silence has veen eliminated. Explains where so much of the new music is boring.
Also reminds me of a conversation with Roger West where he explained about why tubes sound better than transistors by talking about the waveforms and the way transistors and tubes clip the waveform differently. The type of clipping discussed in the srticle would generate tons of high order harmonic distortion, A topic you have written quite a bit about eh Lynn. The shame is that we could have morwe dynamic cintrasts rather than less.
I also wonder about the death wish the music industry has, they do everything possible to make a buck rather than make good recordings and then complain when people download compressed recordings of the compressed music they create, since the sound doesn't really matter.... or people just don't care since the recording is no better than what they hear on MTV or radio, so why pay for the recording?