Optimum Attenuation for HF Driver
This may seems an arcane point, but the sonically optimum way to attenuate the HF compression driver is NOT an L-Pad, but an autoformer with a damping resistor in front of it.
Why? Well, as seen from the Z curves of horns, these things are all over the place. This has a nasty interaction with the crossover, since Zobel compensation is out of the question with multiple impedance peaks - and that's what horns do, thanks to their inherent HOM's.
Multiplying the overall impedance upward with an autoformer obviously attenuates the voltage being fed to the driver. With the simple expedient of a nominal 10-ohm resistor IN PARALLEL with the input side of the autoformer, the Z variations are reduced to almost zero, essentially a fraction of an ohm, which the crossover network appreciates.
There is a further, more subtle benefit. All drivers generate back-EMF's which are sent back to the amplifier. These back-EMF's are filled with resonances and driver distortion, adn we get better sound if there are no obstructions between the back-EMF generator and the amplifier. An L-pad fails here, with its series resistance that is always in the circuit.
The autoformer, by contrast, is transforming impedance, and in a favorable direction from the driver's point of view (it increases the effective damping factor of the amplifier, a good thing). The parallel impedance is helping as well, decreasing the source impedance of the highpass filter - again, as seen from the drivers point-of-view.
The catch is that autoformers with 1 dB taps are most desirable - the replacement ones sold by Klipsch have taps 3 dB apart. Fortunately, the autoformers can be quite small (the Klipsch ones are tiny) since they don't need to handle any LF signal. The small size also improves the HF response of the autoformer, which can go out to 50 kHz without any trouble.
As for vendors, well, I know BudP makes autoformers, and our audio-friend Dave Slagle is a whiz at custom-order iron.
(Quick Note about transformers and autoformers: the turns ratio and voltage ratio is the same, and the impedance ratio is the SQUARE of the turns ratio. So a 6 dB voltage attenuation - 2:1 - is a 4X transformation of impedance.)
This may seems an arcane point, but the sonically optimum way to attenuate the HF compression driver is NOT an L-Pad, but an autoformer with a damping resistor in front of it.
Why? Well, as seen from the Z curves of horns, these things are all over the place. This has a nasty interaction with the crossover, since Zobel compensation is out of the question with multiple impedance peaks - and that's what horns do, thanks to their inherent HOM's.
Multiplying the overall impedance upward with an autoformer obviously attenuates the voltage being fed to the driver. With the simple expedient of a nominal 10-ohm resistor IN PARALLEL with the input side of the autoformer, the Z variations are reduced to almost zero, essentially a fraction of an ohm, which the crossover network appreciates.
There is a further, more subtle benefit. All drivers generate back-EMF's which are sent back to the amplifier. These back-EMF's are filled with resonances and driver distortion, adn we get better sound if there are no obstructions between the back-EMF generator and the amplifier. An L-pad fails here, with its series resistance that is always in the circuit.
The autoformer, by contrast, is transforming impedance, and in a favorable direction from the driver's point of view (it increases the effective damping factor of the amplifier, a good thing). The parallel impedance is helping as well, decreasing the source impedance of the highpass filter - again, as seen from the drivers point-of-view.
The catch is that autoformers with 1 dB taps are most desirable - the replacement ones sold by Klipsch have taps 3 dB apart. Fortunately, the autoformers can be quite small (the Klipsch ones are tiny) since they don't need to handle any LF signal. The small size also improves the HF response of the autoformer, which can go out to 50 kHz without any trouble.
As for vendors, well, I know BudP makes autoformers, and our audio-friend Dave Slagle is a whiz at custom-order iron.
(Quick Note about transformers and autoformers: the turns ratio and voltage ratio is the same, and the impedance ratio is the SQUARE of the turns ratio. So a 6 dB voltage attenuation - 2:1 - is a 4X transformation of impedance.)
There are two medicines mentioned for bone injuries.
1. Symphytum 3x or 6x .
This is what a lot of my friends took for their broken bones. They healed very well. However they took the 30x ( not 3x or 6 x as they were not available) for a whole month , thrice a day . Four tiny pills each time , under the tounge.
2. Ruta 30 . This is to aid the healing process of all bone injuries . That probably means speed it up. Can be taken in conjuction with (1).
Just as a side note. Homeopathic medicines generally work by trigering the immune system . So once you take some pills, you have to let it work. The next dose is taken only when the 'curing' has stopped or slowed down. So traditionally the gap between doses could be long ! However under modern circumstances people take them more often believing ( wrongly ) that more medication will help speed up the process. Any extra medication is simply washed off with body waste.
SOME medicines should not be taken together . This is clearly mentoned in homeopathic books ( Materia Medica ).
Short note on dosage. The number after the medicine name indicates the strength - actually that means dilution . A larger number like say 30 is MORE potent than say 3 or 6 and is a HIGHER dilution.
This is because the more dilute ( highly dispersed molecules ) apparently can penetrate the system more easily. The pills are just milk sugar ( lactose ) as they are neutral carriers. The medicine itself is introduced into it as a dilution in alcohol. Each pill will probably contain a millionth ( or less) of a gram of the actual medicine !
Don't send me flak about this . Hanneman ( who started this system long ago ) was laughed at and chased out of Germany for this theory. He went to France and they benefitted enormously by it ! Homeopathy has been in this part of the world for a VERY long time. The number of users is growing exponentially . In our childhood we ONLY took homeopathic medicines.
All this is not to say that allopathic medicines are not required . Each systems is useful for different applications.
End of sermon.
Edit: Forgot to add a note of the dosage . Why 4 pills ? Are 10 pills better and will that speed up the process ? NO.
If the pills are uniform carriers of the medication , just ONE will do. However since most dilutions are in millionths of a gram , the possibility exists that some pills do not have any medicine in them ! So by taking 4 pills you are pretty sure that at least one of them will carry the medicine .
I'll clarify about the number notation soon. Using it with an X and without does mean different methods of dilution ( and therefore strength). Usually lower numbers ( 3, 6 etc ) are used for acute problems ( just developed problems ) and higher numbers for chronic problems and super dilutions ( 1M ) for very long standing problems. It's a very interesting system of medication. The 1M dilution probably carry's next to nothing --- and has been laughed at for a long time by allopathic doctors . But they seem to work , as time has shown.
1. Symphytum 3x or 6x .
This is what a lot of my friends took for their broken bones. They healed very well. However they took the 30x ( not 3x or 6 x as they were not available) for a whole month , thrice a day . Four tiny pills each time , under the tounge.
2. Ruta 30 . This is to aid the healing process of all bone injuries . That probably means speed it up. Can be taken in conjuction with (1).
Just as a side note. Homeopathic medicines generally work by trigering the immune system . So once you take some pills, you have to let it work. The next dose is taken only when the 'curing' has stopped or slowed down. So traditionally the gap between doses could be long ! However under modern circumstances people take them more often believing ( wrongly ) that more medication will help speed up the process. Any extra medication is simply washed off with body waste.
SOME medicines should not be taken together . This is clearly mentoned in homeopathic books ( Materia Medica ).
Short note on dosage. The number after the medicine name indicates the strength - actually that means dilution . A larger number like say 30 is MORE potent than say 3 or 6 and is a HIGHER dilution.
This is because the more dilute ( highly dispersed molecules ) apparently can penetrate the system more easily. The pills are just milk sugar ( lactose ) as they are neutral carriers. The medicine itself is introduced into it as a dilution in alcohol. Each pill will probably contain a millionth ( or less) of a gram of the actual medicine !
Don't send me flak about this . Hanneman ( who started this system long ago ) was laughed at and chased out of Germany for this theory. He went to France and they benefitted enormously by it ! Homeopathy has been in this part of the world for a VERY long time. The number of users is growing exponentially . In our childhood we ONLY took homeopathic medicines.
All this is not to say that allopathic medicines are not required . Each systems is useful for different applications.
End of sermon.
Edit: Forgot to add a note of the dosage . Why 4 pills ? Are 10 pills better and will that speed up the process ? NO.
If the pills are uniform carriers of the medication , just ONE will do. However since most dilutions are in millionths of a gram , the possibility exists that some pills do not have any medicine in them ! So by taking 4 pills you are pretty sure that at least one of them will carry the medicine .
I'll clarify about the number notation soon. Using it with an X and without does mean different methods of dilution ( and therefore strength). Usually lower numbers ( 3, 6 etc ) are used for acute problems ( just developed problems ) and higher numbers for chronic problems and super dilutions ( 1M ) for very long standing problems. It's a very interesting system of medication. The 1M dilution probably carry's next to nothing --- and has been laughed at for a long time by allopathic doctors . But they seem to work , as time has shown.
Thanks for the Info
Many thanks for the info. I'll be visiting Boulder in the next few days and will check this out. I also agree with comments about flamers - if we think the ABX people are bad, just get the professional debunkers going about homeopaths!!!
One of the good things about growing up in Japan and Hong Kong was the matter-of-fact attitude towards the metaphysical world. If it worked, end of story. A vice-president of Sony or a Hong Kong businessman goes to the Shinto or Taoist temple for weddings, a Buddhist temple for funerals, a Feng Shui master to improve their business, an acupuncturist for an annoying chronic condition, a Harvard Business School grad for marketing information on how to enter the US market, and an MIT grad on how to design a new gizmo. All in a day's work in Hong Kong, Tokyo, or Osaka.
Many thanks for the info. I'll be visiting Boulder in the next few days and will check this out. I also agree with comments about flamers - if we think the ABX people are bad, just get the professional debunkers going about homeopaths!!!
One of the good things about growing up in Japan and Hong Kong was the matter-of-fact attitude towards the metaphysical world. If it worked, end of story. A vice-president of Sony or a Hong Kong businessman goes to the Shinto or Taoist temple for weddings, a Buddhist temple for funerals, a Feng Shui master to improve their business, an acupuncturist for an annoying chronic condition, a Harvard Business School grad for marketing information on how to enter the US market, and an MIT grad on how to design a new gizmo. All in a day's work in Hong Kong, Tokyo, or Osaka.
One rather cool thing about the Eminence based coax drivers is that you can thread on any 1" driver I believe. Some would require an adaptor from bolted to threaded, but these are available. While I constantly read that the Eminence horn drivers are quite good for the money, that kind of implies that there are better out there for more money..
I really am not in the know, but it seems that 18 Sound stuff is mentioned with reverence these days... and it doesn't seem as pricey as some...
Here's Euro Prices...
http://www.kcking.sk/18S_EUR06.PDF
This one:
http://www.eighteensound.com/index.aspx?mainMenu=view_product&pid=189
is only $150 at usspeaker.com
That curve is crazyflat. Smoothed, but still great, no?
This one , the The ND1020:http://www.eighteensound.com/index.aspx?mainMenu=view_product&pid=191
looks about the same price as the above. Not quite as flat, but actually very close, and is their only horn driver that says: " ...has been designed for use in situations where the highest possible sound quality is required." and doesn't use a titanium diaphram-which is often a good thing from what I hear people say.
Here is the 15" coax. Not cheap but maybe not bad value..
http://www.eighteensound.com/index.aspx?mainMenu=view_product&pid=247
http://www.loudspeakersplus.com/html/18sound_co-ax.html
Both coax's use a 1.4 inch HF driver, the 1480- a $400 horn driver. With a bigger driver like this the crossover can be lower...
PS:
Don't we need one of these for the bass augmentation?
http://www.loudspeakersplus.com/images/21LW1400.pdf
Only $500 and 99 dB Pretty much a circular planar speaker!
OK, this can go on all day so I'll stop babbling. I suppose one could e-mail each manufacturer and ask which of their drivers is most musical, and that we don't care about : power handling, weather resistance, or efficiency- almost any pro driver has more of these parameters than we would ever need..
I really am not in the know, but it seems that 18 Sound stuff is mentioned with reverence these days... and it doesn't seem as pricey as some...
Here's Euro Prices...
http://www.kcking.sk/18S_EUR06.PDF
This one:
http://www.eighteensound.com/index.aspx?mainMenu=view_product&pid=189
is only $150 at usspeaker.com
That curve is crazyflat. Smoothed, but still great, no?
This one , the The ND1020:http://www.eighteensound.com/index.aspx?mainMenu=view_product&pid=191
looks about the same price as the above. Not quite as flat, but actually very close, and is their only horn driver that says: " ...has been designed for use in situations where the highest possible sound quality is required." and doesn't use a titanium diaphram-which is often a good thing from what I hear people say.
Here is the 15" coax. Not cheap but maybe not bad value..
http://www.eighteensound.com/index.aspx?mainMenu=view_product&pid=247
http://www.loudspeakersplus.com/html/18sound_co-ax.html
Both coax's use a 1.4 inch HF driver, the 1480- a $400 horn driver. With a bigger driver like this the crossover can be lower...
PS:
Don't we need one of these for the bass augmentation?
http://www.loudspeakersplus.com/images/21LW1400.pdf
Only $500 and 99 dB Pretty much a circular planar speaker!
OK, this can go on all day so I'll stop babbling. I suppose one could e-mail each manufacturer and ask which of their drivers is most musical, and that we don't care about : power handling, weather resistance, or efficiency- almost any pro driver has more of these parameters than we would ever need..
Lynn Olson said:
Well.. Not exactly. Remember this is a compression driver, and to an extent a more normal "flat" impedance/mass controlled diaphram is not quite the same standard..
Better to have a linear decay representation in a good horn eh?
(thanks to Paul W!)
http://www.diyaudio.com/forums/showthread.php?s=&threadid=90804&perpage=10&pagenumber=13
So, 1.2 kHz is the problem area, still - that problem area is down almost 20 db before .5 milsec's. Not too shabby at 1.2 kHz when compared to many drivers at this freq..
(..and even then, how much of this is an effect of the horn?)btw, DON'T look at his diy ribbon decay - it will make you
Not bad, $625 for the 15CX1000, at http://www.loudspeakersplus.com/html/18sound_co-ax.html
Some of the cheapskates might scream, but look at all the driver you get! A 15-inch bass driver with the most advanced distortion-reduction magnetic system, and a 1.4-inch pro-quality compression driver of quite modern design. Compare THAT to low-efficiency audiophile alternatives, or trying to find matched pairs of sixty-year-old Altecs or Tannoys - at astronomical prices.
I like what I see here.
Some of the cheapskates might scream, but look at all the driver you get! A 15-inch bass driver with the most advanced distortion-reduction magnetic system, and a 1.4-inch pro-quality compression driver of quite modern design. Compare THAT to low-efficiency audiophile alternatives, or trying to find matched pairs of sixty-year-old Altecs or Tannoys - at astronomical prices.
I like what I see here.
I have looked at a lot of HF driver curves over the last couple of years, and those 18 Sound ones are freakishly flat. Flatter than many conventional hifi tweeters..It makes you want to use their 60x80 horn (which is how the driver is measured) instead of the coax.. I'm too ignorant to understand any tricks they might be pulling though.. Of course, flat response isn't everything, that's why I'm intrigued by that driver they say is for the highest quality- which isn't available in the USA apparently
In the 15" coax they are putting their money into the HF driver which is probably the right thing to do.. $600 might be half of what the Hemp Audio 15 costs...
Don´t forget the B&C coaxials:
15"
http://www.bcspeakers.com/index.php...rizione=32&prodotto=59&id_descrizione_prod=55
12"
http://www.bcspeakers.com/index.php...rizione=32&prodotto=58&id_descrizione_prod=75
http://www.bcspeakers.com/index.php...rizione=32&prodotto=58&id_descrizione_prod=75
B&C, 18Sound and BMS (the last one for compression drivers) seem to be the top PA driver producers in Europe now.
I understand the the exit of the CD and the transition to the "cone horn" are critical. Are this really well addressed in the coax designs?
What about the cone movement, how would 2-4mm p-p movement affect the cone horn fot the tweeter around XO?
FWIW, E. Geddes comments coaxes intoduce more problems than what they solve.
The 18Sound constand directivity waveguides are very well regarded. According to a distributor in Europe that carries PAudio and B&C says the PAudio coaxes are no match to the B&C, and suggest non-coaxes for better results.
15"
http://www.bcspeakers.com/index.php...rizione=32&prodotto=59&id_descrizione_prod=55
12"
http://www.bcspeakers.com/index.php...rizione=32&prodotto=58&id_descrizione_prod=75
http://www.bcspeakers.com/index.php...rizione=32&prodotto=58&id_descrizione_prod=75
B&C, 18Sound and BMS (the last one for compression drivers) seem to be the top PA driver producers in Europe now.
I understand the the exit of the CD and the transition to the "cone horn" are critical. Are this really well addressed in the coax designs?
What about the cone movement, how would 2-4mm p-p movement affect the cone horn fot the tweeter around XO?
FWIW, E. Geddes comments coaxes intoduce more problems than what they solve.
The 18Sound constand directivity waveguides are very well regarded. According to a distributor in Europe that carries PAudio and B&C says the PAudio coaxes are no match to the B&C, and suggest non-coaxes for better results.
ScottG said:Actually their most advanced distortion reduction system is reserved for the "Active Impedance" models. (..none of which are coaxs unfortunately.) Well - thats distortion reduction for the larger driver.. not the compression driver.
I think you are talking about this one:
http://www.eighteensound.com/index.aspx?mainMenu=view_product&pid=240
18 Sound also has a new 15" coming.
http://www.eighteensound.com/index.aspx?mainMenu=view_product&pid=265
Hi all,
Just a note that you want to be a bit careful with how the step down or step up transformer is put together and what it's goals are.
When you are running signal level currents and voltages an auto former that is essentially a tapped winding is ok. You can do better but not breathtakingly so, because the irregularities introduced by "auto forming" are small at low power levels.
When you get into relatively high current situations, with modest voltages, as we are talking about in speaker drive applications a true auto former is going to introduce E Field coupling problems that are quite audible. The E Field requires large surfaces and true dielectrics to work properly and so you end up with interlayer's of primary and secondary windings and perform the auto former hook up outside of the coil.
Even then the difference between an auto former and a true isolation transformer is audible. Not grossly audible, and without two units to compare you will not be dissatisfied with that level of auto former performance.
Yes I am aware that it is the B Field that current creates. The E Field is the precursor to all B Field current flow, for every vector change in the fields, describing the transformation process. Without enough coupling surface for a weak E Field, as with these relative low voltage high current devices, you will loose transient attack and micro dynamic emphasis color and coherency. The signal will become thin and sharp sounding, rather than the rather dense and liquid sound we do prefer, when we can get it.
One dB steps will be a pain. They can be done, but you should come up with a range across which you want to step, of probably 6 steps. This is four primaries driving three secondary zones with a two layer tapped secondary winding per zone.
This scheme will be a bit more expensive than you might expect, but the sonic benefits are worth it.
A number of years ago we were supplying OPT's to Audio Prism and had helped them get their performance level up to the point that they were very busy and caught the eye of one M Levinson.
He was attempting to bring a huge ribbon and Dynaudio system to market and the ribbons required an impedance matching transformer. The unit they had was an auto former, built on a 1.5" X 2 " center core dimension with #9 awg wire wound first and the non current additive portion of the winding wound over that, in very typical auto former construction. Basically a welding transformer, and it sounded like it.
Mr. Levinson eventually replaced every one of these devices, in the field, with a properly designed impedance matching transformer. He did insist we stay with the original designers concept (Bo Bingston) of using an auto former, rather than a fully isolated design, but he was quite aware of the sonic differences even with both devices utilizing proper coupling structures.
Just so you know that you can defeat all of your good work by being thrifty and ignorant here, in particular.
Bud
Just a note that you want to be a bit careful with how the step down or step up transformer is put together and what it's goals are.
When you are running signal level currents and voltages an auto former that is essentially a tapped winding is ok. You can do better but not breathtakingly so, because the irregularities introduced by "auto forming" are small at low power levels.
When you get into relatively high current situations, with modest voltages, as we are talking about in speaker drive applications a true auto former is going to introduce E Field coupling problems that are quite audible. The E Field requires large surfaces and true dielectrics to work properly and so you end up with interlayer's of primary and secondary windings and perform the auto former hook up outside of the coil.
Even then the difference between an auto former and a true isolation transformer is audible. Not grossly audible, and without two units to compare you will not be dissatisfied with that level of auto former performance.
Yes I am aware that it is the B Field that current creates. The E Field is the precursor to all B Field current flow, for every vector change in the fields, describing the transformation process. Without enough coupling surface for a weak E Field, as with these relative low voltage high current devices, you will loose transient attack and micro dynamic emphasis color and coherency. The signal will become thin and sharp sounding, rather than the rather dense and liquid sound we do prefer, when we can get it.
One dB steps will be a pain. They can be done, but you should come up with a range across which you want to step, of probably 6 steps. This is four primaries driving three secondary zones with a two layer tapped secondary winding per zone.
This scheme will be a bit more expensive than you might expect, but the sonic benefits are worth it.
A number of years ago we were supplying OPT's to Audio Prism and had helped them get their performance level up to the point that they were very busy and caught the eye of one M Levinson.
He was attempting to bring a huge ribbon and Dynaudio system to market and the ribbons required an impedance matching transformer. The unit they had was an auto former, built on a 1.5" X 2 " center core dimension with #9 awg wire wound first and the non current additive portion of the winding wound over that, in very typical auto former construction. Basically a welding transformer, and it sounded like it.
Mr. Levinson eventually replaced every one of these devices, in the field, with a properly designed impedance matching transformer. He did insist we stay with the original designers concept (Bo Bingston) of using an auto former, rather than a fully isolated design, but he was quite aware of the sonic differences even with both devices utilizing proper coupling structures.
Just so you know that you can defeat all of your good work by being thrifty and ignorant here, in particular.
Bud
agent.5 said:
Well, the 12NDA520 bass driver, XT120 horn, and ND1030 compression drivers certainly have smooth curves - obviously more smoothing is applied than B&C is using! The B&C's have nulls with 12 dB peak-to-dip ratios, nothing you'd attempt to equalize, considering how narrow they are. But they could be there in the 18Sound drivers too, although they certainly must be well concealed. Now, I know from measuring Fostex drivers that the published curves bore no resemblance at all to what I saw on MLSSA on a flat baffle - my unsmoothed MLSSA measurements showed 10 to 15 dB peak-to-dip ratios, nothing like the published curves at all.
By contrast, the drivers in the Ariel are much smoother, none of that narrowband notch-and-peak structure that is so common in high-efficiency drivers. HE enthusiasts ignore, tolerate, or accept the rough response, but it's nothing I want in any speaker I design. I agree with the previous posts that the prevalence of smoothing - which is completely understandable considering the rough respone of much of pro stuff - makes it very difficult to discern what any of these drivers are going to really do.
If the bass is measured in a box instead of a flat baffle, OK, I know what that does - makes peak-dips in what would otherwise be the completely flat piston region. A normally operating 12" driver mounted on a large flat baffle should be flat through 500 to 800 Hz, and if ripples appear, that can be put down to spider resonance (which typically looks like a very narrow ripple on the impedance and freq resp curves). Bigger lumps and dips in the 500 Hz to 1.5 kHz region can be ascribed to cabinet instead of flat-baffle measurements, so these can be safely set aside.
It's the stuff from 1.5 khz on up that concerns me, especially if it is violent enough to appear both the impedance and freq resp curves. These are resonances so severe that direct mechanical correction (fixing the problem at the source) is more fruitful than equalization - since the really big ones appear in the time, frequency, impedance, and polar-pattern domains.
The trouble region for all of these coaxes is obviously in the 1 to 5 kHz region, with many of them producing raw-driver response curves that are beyond the range of successful equalization except in the grossest, PA-application sense. The reputation for coloration is well deserved - these are frankly quite poor response curves, in the same quality range as 6 x 9 car speakers.
The curves from Radian and 18Sound look the best, except we know that lots of smoothing has been applied, obscuring the true picture. Hmmm. We do have the reputation of the driver to consider - with the problem, for me, that lots of people have high regard for Altec and JBL too, and I can't stand that kind of sound.
I didn't like JBL's when I sold them retail in 1972, and I don't like them any better now, some 35 years later. So acceptance in the pro world is kind of problematic - they're listening for different qualities than I am, so recommendation may or may not mean anything. I've never liked speakers that sound harsh in the 1 to 5 kHz region - it means most of classical music is unlistenable - but that is a popular sound in the pro world. The JBL L100 was specifically developed to mimic the midrange colorations of the Altec Duplex, for example, so it's not a fairy tale - and I sold a lot of L100's in Los Angeles, where Altec and JBL reign supreme.
Can coaxes be made that are smooth in the critical midrance region? Well, the 12 and 15-inch paper-cone Tannoys succeeded. The extreme HF may not have been all that extended by modern standards, but the midrange was smooth, much smoother than the Altec Duplex, which became the prototype for "coax sound" to this day. It's taken me quite a while to realize that many people, audiophiles and pros alike, prefer an aggressive, forward, and peaky midrange - back in my retail days, it was a very popular sound - maybe for the simple reason it sounds LOUD and in-your-face.
Although the theoretical benefits of a coax - particularly in an open baffle - are obvious, if the midrange colorations cannot be corrected, a 12" driver like the 18Sound 12NDA520 coupled with either a big prosound ribbon or the 18Sound horn/driver might be a better answer. From where I sit, the crossover sure looks a lot easier, and that wins points for me, since I don't have to throw so many exotic and extremely expensive audiophile capacitors at it. (Inductors I can live with, it's those doggone caps that are critical to the sound, and the more transparent the speaker, the more obvious the cap-coloration gets.)
Re: Autoformers and Transformers
I defer to the experience of BudP, who knows his magnetics. Although I used the word "autoformer" loosely, I was including both autoformers and true transformers to refer to all forms of magnetic level control at typical speaker impedances and voltage levels.
I'm guessing that 4 or 5 steps would be plenty, with a nominal attenuation of 6 to 10 dB with the attenuation switch in the "center" position. It would depend on the horn and how much EQ it needed - some of these things have quite a bit of shelving EQ needed at high frequencies.
I defer to the experience of BudP, who knows his magnetics. Although I used the word "autoformer" loosely, I was including both autoformers and true transformers to refer to all forms of magnetic level control at typical speaker impedances and voltage levels.
I'm guessing that 4 or 5 steps would be plenty, with a nominal attenuation of 6 to 10 dB with the attenuation switch in the "center" position. It would depend on the horn and how much EQ it needed - some of these things have quite a bit of shelving EQ needed at high frequencies.
salas said:
I completely agree. From the looks of some of these drivers, there will be a 2nd-order network followed by a shelving network so the energy at 4 kHz and 16 kHz are reasonably level - and then followed by an L-Pad (for development purposes) or an auto/transformer. Not having to use a notch filter would be a big plus in terms of network complexity.
Lynn Olson said:
Maybe we can request 18 sound to custom make us some coaxial
12NDA520
+
NSD1480
http://www.eighteensound.com/index.aspx?mainMenu=view_product&pid=182
Lynn Olson said:
While cone profile plays its part, getting a fairly smooth response without severe break-up at the top end largely requires a smaller VC, and that works against most pro designs that strive for low thermal compression and lots of space for a coax tweeter.
IMO you should be pursuing a different route more common to the fullrange crowd. i.e. a broadband midrange and "filler" drivers for both ends of the freq..
I particularly like the idea of the B&O filler driver crossover - it has these things going for it:
1. Steep filter at crosspoint for both the woofer and the tweeter, minimizing their greatest weaknesses and allowing for more "exotic" solutions.
2. With a fair bit of artistic implementation (with driver selection and the baffle), the low pass filter transfer of the woofer could be incorporated into some of the open baffle loss compensation.
3. The filler driver IS your point source in the critical midband, and if chosen well can have excellent measured performance.
4. The filter requires all drivers on a flat baffle, thus making physical alignment a "breeze".
5. The design (with careful selection of the filler driver) can exhibit excellent dispersion character at higher freq.s.
6. Its a time perfect design.
The downside is that the filler driver needs to be VERY efficient (aproximatly 4-6 db more than the average depending on Gamma), and its also more difficult to achieve a nominally resistive impedance profile (..which again requires more thought into driver selection).
Thoughts on filler drivers:
Looking for eff., bandwidth with excursion, AND a small diameter for better dispersion character is TOUGH. The real alternative here is with a larger diameter driver and asking a bit more from the tweeter and a bit less from the woofer. Additionally IF you want to maintain open baffle operation for this driver you will need something with at least a modest amount of excursion. Another alternative is to "swing" in the other direction by shifting the passband of the filler driver higher in freq..
If the spec.s don't lie, (they certainly do upon modeling), then the MaxFidelity PR4Neo/8, might well work.
http://www.solen.ca/
Increasing filler driver sd (which will require more from the tweeter), could be found in several drivers both pro and "fullrange".
If you want something with an even smaller sd, OR you want greater power handling, consider a small line array that meets the restriction for point source character for the respective listening distance and is down far enough in spl's beyond the lowpass crosspoint so that any combing doesn't become intrusive.
LOTs of possibilities.
Well, just a suggestion.
Hmm, I'm a little cool on the filler driver for these reasons:
1) The phase angle between all three drivers is 90 degrees, or phase quadrature. This is something skipped over in the filler-drive advocacy groups. In more detail, if the tweeter is at +90 degrees, the filler is at 0 degrees, and the woofer is -90 degrees. In other words, with the filler driver shut off, there is an infinitely deep null created between the HF driver and the woofer.
Moreover, since the filler-drive physically sits between the HF and woofer, the spacing requirement is pretty considerable, thus creating a quite narrow polar pattern vertically. That's why most commercial implementations of filler had relatively low crossovers - a conventional crossover at, say, 3 kHz would end up with a polar pattern maybe 10~15 degrees wide at the listening position, leading to annoying venetian-blind effects as you sit down and stand up.
2) What does a filler driver actually do? It avoids the necessity for phase inversion in what would otherwise be a perfectly conventional 2nd-order crossover. Seriously, that's it. In return for a much narrower polar pattern, you get linear phase response.
3A) Here is the most serious problem of all, which goes unaddressed by the FD advocates. Direct-radiator driver excursion increases at a rate of 12 dB/octave as a result of the constant-acceleration nature of direct-radiators. This means any 1st-order highpass crossover (as used by the FD) actually has increasing excursion between the nominal crossover frequency and the fundamental resonance of the driver. (Below which the acoustical rolloff increases from 6 dB/oct to 18 dB/oct, which increases the phase angle between the drivers from the 90-degree design-center.)
3B) The increasing excursion from the filler driver actually results in a system that has more IM distortion than the simple 2-driver 12 dB/octave system it replaces. Remember, the filler driver is doing nothing more than offsetting the acoustical cancellation created by the HF drivers and woofer being out of phase with each other. It sounds bizarre, but simply reversing the phase of those two drivers makes the filler driver (and its distortion) unnecessary! Again, the filler driver is not a midrange, despite appearances: it is precisely "filling in" (and nothing more) the null created by having two drivers connected out-of-phase.
4) Yes, it is linear phase, but at the expense of a narrow vertical polar pattern, and having three drivers working at cross-purposes to each other. Wasting acoustical energy by deliberately creating cancellation nulls bothers me at an intuitive level. And without the cancellation null, there's nothing for the filler driver to do, since it overlaps what would otherwise by a normal crossover region with 12 dB/oct slopes on each side.
5) If linear phase is important - and it is to some designers - I've found you can "cheat" by using low-Q (0.5) 2nd-order crossovers and using increased driver offsets. You don't get perfect square waves, but they're not too bad. A system I designed in 1979 for Audionics did just this, using the Audax 6.5" Bextrene driver and an Audax 1" dome in a low-diffraction satellite-style cabinet and a dedicated push-pull subwoofer.
Sorry to rank on the FD so hard, but this is one of the bad ideas that just won't go away. Every FD speaker I've heard sounds quite phasey in the crossover region, and steep phase angles and a narrow polar pattern is why. The ear - OK, my ears at least - are much more sensitive to inter-driver phase angles than phase distortion in the absolute sense, since inter-driver phase directly controls polar pattern shapes at the crossover frequency. Phase distortion in the absolute sense affects timbre - at lower frequencies - but of course has no effect on polar patterns.
Returning to the 18Sound discussion, the performance of the previously-mentioned 12" drivers and the top-of-the-line 18Sound horn/driver sets is almost making me re-think the whole coax thing. Simple crossovers combined with inherently smooth driver response is a rare virtue in the high-end biz, and even more rare in the high-efficiency world.
An 18Sound top end, combined with 2 or 4 Alnico & Ceramic Tone Tubby's supporting the bass module, could make a stunningly good 99 dB/metre efficient dipole loudspeaker. I'll throw it out to you guys - of all the 18Sound horn/CD combos, which are the best? (Most advanced CD, most advanced horn with the smoothest response, you get the idea.)
1) The phase angle between all three drivers is 90 degrees, or phase quadrature. This is something skipped over in the filler-drive advocacy groups. In more detail, if the tweeter is at +90 degrees, the filler is at 0 degrees, and the woofer is -90 degrees. In other words, with the filler driver shut off, there is an infinitely deep null created between the HF driver and the woofer.
Moreover, since the filler-drive physically sits between the HF and woofer, the spacing requirement is pretty considerable, thus creating a quite narrow polar pattern vertically. That's why most commercial implementations of filler had relatively low crossovers - a conventional crossover at, say, 3 kHz would end up with a polar pattern maybe 10~15 degrees wide at the listening position, leading to annoying venetian-blind effects as you sit down and stand up.
2) What does a filler driver actually do? It avoids the necessity for phase inversion in what would otherwise be a perfectly conventional 2nd-order crossover. Seriously, that's it. In return for a much narrower polar pattern, you get linear phase response.
3A) Here is the most serious problem of all, which goes unaddressed by the FD advocates. Direct-radiator driver excursion increases at a rate of 12 dB/octave as a result of the constant-acceleration nature of direct-radiators. This means any 1st-order highpass crossover (as used by the FD) actually has increasing excursion between the nominal crossover frequency and the fundamental resonance of the driver. (Below which the acoustical rolloff increases from 6 dB/oct to 18 dB/oct, which increases the phase angle between the drivers from the 90-degree design-center.)
3B) The increasing excursion from the filler driver actually results in a system that has more IM distortion than the simple 2-driver 12 dB/octave system it replaces. Remember, the filler driver is doing nothing more than offsetting the acoustical cancellation created by the HF drivers and woofer being out of phase with each other. It sounds bizarre, but simply reversing the phase of those two drivers makes the filler driver (and its distortion) unnecessary! Again, the filler driver is not a midrange, despite appearances: it is precisely "filling in" (and nothing more) the null created by having two drivers connected out-of-phase.
4) Yes, it is linear phase, but at the expense of a narrow vertical polar pattern, and having three drivers working at cross-purposes to each other. Wasting acoustical energy by deliberately creating cancellation nulls bothers me at an intuitive level. And without the cancellation null, there's nothing for the filler driver to do, since it overlaps what would otherwise by a normal crossover region with 12 dB/oct slopes on each side.
5) If linear phase is important - and it is to some designers - I've found you can "cheat" by using low-Q (0.5) 2nd-order crossovers and using increased driver offsets. You don't get perfect square waves, but they're not too bad. A system I designed in 1979 for Audionics did just this, using the Audax 6.5" Bextrene driver and an Audax 1" dome in a low-diffraction satellite-style cabinet and a dedicated push-pull subwoofer.
Sorry to rank on the FD so hard, but this is one of the bad ideas that just won't go away. Every FD speaker I've heard sounds quite phasey in the crossover region, and steep phase angles and a narrow polar pattern is why. The ear - OK, my ears at least - are much more sensitive to inter-driver phase angles than phase distortion in the absolute sense, since inter-driver phase directly controls polar pattern shapes at the crossover frequency. Phase distortion in the absolute sense affects timbre - at lower frequencies - but of course has no effect on polar patterns.
Returning to the 18Sound discussion, the performance of the previously-mentioned 12" drivers and the top-of-the-line 18Sound horn/driver sets is almost making me re-think the whole coax thing. Simple crossovers combined with inherently smooth driver response is a rare virtue in the high-end biz, and even more rare in the high-efficiency world.
An 18Sound top end, combined with 2 or 4 Alnico & Ceramic Tone Tubby's supporting the bass module, could make a stunningly good 99 dB/metre efficient dipole loudspeaker. I'll throw it out to you guys - of all the 18Sound horn/CD combos, which are the best? (Most advanced CD, most advanced horn with the smoothest response, you get the idea.)
Thats cool ("ranking")on the FD approach - though I think many of the problems can be compensated for, I did not consider vertical problems!
More importantly, I've never heard one. ..And if it sounds wrong, then it IS wrong.
(Note: I had the same problems with Dunlavey's designs, despite how well they measured on axis.)
More importantly, I've never heard one. ..And if it sounds wrong, then it IS wrong.
(Note: I had the same problems with Dunlavey's designs, despite how well they measured on axis.)