the ancient urge to build a dual-woofer three-way

Pat McGinty

Senior Member
Huh? Just our silly tradition - we don't name 'em until they get the thumbs up and are a "go" for production. That's part-and-parcel of our disposable prototype philosophy; do not get emotionally involved with something that you MUST be able to kill without flinching. So, until that time they carry a deliberately unusable moniker, usually humorous. We did almost violate that regime just once, along time back, when we came perilously close to bringing the Titmouse to market. A last minute attack of sobriety and discretion resulted in Vireo.


vireo logo.jpg


You know guys, you can't save money on speakers. If you buy middling ones you'll just torture yourself at the back of your mind while trying to enjoy them. The Poor Man Always Pays Twice. First rate transducers cost plenty, but why not own 'em early and wear 'em out?

That idea with an Illuminator and a Mundorf AMT makes for a killer "bedroom speaker". Add, say, the 2x125W N-Core plate and a competent enclosure and you certainly do not have a cheap speaker (actually, it's not a speaker, it's a complete system sans streamer). I gotta add that a dipole Mundorf paints a paanoooraaamaaa all over the room.

But, hey, I do have the transducers on hand and need to place an amp order with Hypex, anyway, for the 1KW plates we need for the big quad pump flame throwers.....so step right up.
 

Prime Minister

Site Owner
Staff member
Who
You know guys, you can't save money on speakers. If you buy middling ones you'll just torture yourself at the back of your mind while trying to enjoy them. The Poor Man Always Pays Twice. First rate transducers cost plenty, but why not own 'em early and wear 'em out?

That idea with an Illuminator and a Mundorf AMT makes for a killer "bedroom speaker". Add, say, the 2x125W N-Core plate and a competent enclosure and you certainly do not have a cheap speaker (actually, it's not a speaker, it's a complete system sans streamer). I gotta add that a dipole Mundorf paints a paanoooraaamaaa all over the room.

But, hey, I do have the transducers on hand and need to place an amp order with Hypex, anyway, for the 1KW plates we need for the big quad pump flame throwers.....so step right up.
Who said anything about a small speaker? :)

I was thinking something like a very efficient, 10 inch driver. The AMT would be a lovely tweeter to match up with it. Have you ever messed with something like that?
 

Ernie

Activated
Does the 10" extend high enough to mate with the AMT, which likes a 2K xo, using conventional passive xo devices? Does the DSP "infinite slope", figuratively speaking, mean that you can go lower with the tweeter, and higher with the woofer, beyond conventionally-adopted limits?
 

Ernie

Activated
I've heard some different small speakers with AMT tweeters, and always been impressed with the imaging. These were not dipoles, and they sounded quite happy at low and high volumes.
 

Prime Minister

Site Owner
Staff member
Does the 10" extend high enough to mate with the AMT, which likes a 2K xo, using conventional passive xo devices? Does the DSP "infinite slope", figuratively speaking, mean that you can go lower with the tweeter, and higher with the woofer, beyond conventionally-adopted limits?
I'm guessing you are not asking those questions of me. :)
 

Ernie

Activated
Only if you can answer them. Actually, my question was in reference to the Scanspeak 10" woofers, but if you were trying for a 10" 2-way, then I guess you'd need something with specs more like the Eminence 10s but with a more refined character, and voice. I built a QWTL for a friend of mine, who wanted speakers for in his garage, to provide sound for his back yard. Beta 10CX, complete with the Eminence compression driver, crossed at 3.5KHz. He drove them with a G-33000, and frequently explored the top of the volume range. The speakers were standing in an open overhead door, facing the yard. The empty garage made quite the resonant chamber. Serbian mp3s. It was, um, exciting, from a distance.
 

Ernie

Activated
As I ruminate on this, can you get a speaker that will perform well, wrt detail and volume, at low volumes, and still run up to your H2H levels? It seems counter-intuitive, as you'd need something pretty low in mass to run well at low volumes, but how does it fight break-up at high SPLs?

I'm just rambling, wool-gathering, as it were.
 

Pat McGinty

Senior Member
I've used the 5"Mundorf down to the mid teens with my normal (!) working amplitudes. Naturally, you can lower the corner F along with your output demands. So a guy could have one Preset for low level and another doing more work.

Also, I have beaten the crap out of the Mundorfs, mercilessly, and have yet to destroy one. Unlike a dome tweeter's coil, the trace is quite long and spread out so the heat is dissipated across a large, well ventilated area, plus the diaphragm is a heat resistant polymer, kapton I think. The excursion limit is when adjacent folds collide - which doesn't hurt anything but does give you an obvious report. So, yeah, you won't pop 'em so go ahead and take all the bandwidth you can get away with.

Steep filters by DSP can be used to get better quality trebles than passives because you can brick-wall exclude LF that would have been raising excursion and temp. Not to mention the nasty signal deterioration that occurs in LCR circuits.

One other thought: if you're asking for very little output there's no reason not to EQ for full bandwidth since you'd likely have the woofer displacement to pull it off. Maybe dial in a touch of Fletcher Munson EQ while you're at it.

Maybe I oughta dig out a couple of five inchers, set 'em on top of my Kingfishers, and report back? It's not the right enclosure for work into the mids, but might give a clue.
 

Prime Minister

Site Owner
Staff member
I've used the 5"Mundorf down to the mid teens with my normal (!) working amplitudes. Naturally, you can lower the corner F along with your output demands. So a guy could have one Preset for low level and another doing more work.

Also, I have beaten the crap out of the Mundorfs, mercilessly, and have yet to destroy one. Unlike a dome tweeter's coil, the trace is quite long and spread out so the heat is dissipated across a large, well ventilated area, plus the diaphragm is a heat resistant polymer, kapton I think. The excursion limit is when adjacent folds collide - which doesn't hurt anything but does give you an obvious report. So, yeah, you won't pop 'em so go ahead and take all the bandwidth you can get away with.

Steep filters by DSP can be used to get better quality trebles than passives because you can brick-wall exclude LF that would have been raising excursion and temp. Not to mention the nasty signal deterioration that occurs in LCR circuits.

One other thought: if you're asking for very little output there's no reason not to EQ for full bandwidth since you'd likely have the woofer displacement to pull it off. Maybe dial in a touch of Fletcher Munson EQ while you're at it.

Maybe I oughta dig out a couple of five inchers, set 'em on top of my Kingfishers, and report back? It's not the right enclosure for work into the mids, but might give a clue.
Not to throw stones at other designers or designs, but I don't get the impression that you are crazy about large (say 10 inch or greater), high sensitivity woofers. Have you ever played with setups like that? Ever tried very high sensitivity woofers?
 
Since this is about to get buried in another speaker thread, here is my summary of why I think what you are doing is so amazing/the best method to get best potential sound.

Why “digispeakers”?

1. The elimination of (as far as Im concerned now that we have the technology) many unnecessary metal/metal interfaces (interconnect and speaker cable interfaces) as well as interconnects themselves (in the case of digital sound sources)and speaker cables themselves.

“Digispeakers” make what is written below irrelevant. Lets take the analog out of a DAC in a traditional stereo as an example. There are going to be a minimum of 8 metal/metal interfaces that the electrical signal (4 for each channel) and a maximum of....Pre-amp/amp anyone? EQ anyone? yadada the list can go on and the interfaces stack up the more complex the system. Why not crunch numbers instead and BYE BYE metal/metal interfaces and resulting sound degradation. Don't forget the speaker cable interfaces as well. CYA!

With no metal/metal interfaces, forget about the low (probably <<2%) “real area of contact” in interconnect and speaker connections like bananas. For speaker I prefer bare wire then spades I can crank down (fresh gouges in metal are good from an electrical conductivity point of view, though even what was once was true (no oxides, sulfides, hydrocarbons, etc) metal/metal contact will immediately started becoming contaminated the second after they are created, unless the environment is an ultra high vacuum (good luck hearing or living in one of those :)). Those RCA and XLR and banana, etc contact surfaces may look nice and smooth and shiny but get them under a microscope and the surfaces are huuuge peaks and valleys (I used to do atomic force microscopy as well as less sensitive mechanical surface metrology using diamond styluses). And this doesn't even take into account how precisely they are machined. Or how precisely the male surface mates to the female surface (different brands, manufacturing tolerances, total indicated runout(how off they are from being perfectly circular), etc). Its mostly air surface area (probably >>98%) between RCA/XLR/speaker connections, etc. This is where micro-arcing can occur. One reason for contact enhancers such as Cardas TC-2 Contact Conditioner (they also have a cleaner).

"Cardas Contact Conditioner cleans and improves all electrical contact surfaces. Apply a trace amount and burnish into the surface with a soft cloth. After cleaning, a second trace amount is applied and rubbed into the metal to fill microscopic irregularities. This creates a smoother surface for improved signal transfer at the contact interface. It only takes a small amount to enhance the performance of a connection. To maintain performance, reapply every thirty to sixty days. Available in 3 ml bottles."

Or Caig Deoxit Gold Contact Enhancer. (no doubt there are many others as well, like Tweak from way back when, etc)

  • Enhances, protects, lubricates, and maintains optimum signal quality
  • Reduces intermittent connections, arcing, and RFI as well as wear and abrasion
  • Provides long-lasting protection from oxidation on plated surfaces
While I have never done surface metrology on RCA or XLR (or whatever metal/metal stereo interface) surfaces, I can say with 99.99% certainty the distances in the air gap between male/female metal/metal interfaces will certainly be up to the micrometer (hence micro-arcing) range. They are hardly polished like the wafers used for making chips are. There are some high dollar esoteric RCA plugs that have a single spherical contact for the outer connection (the rest of the "outer circular RCA plug" is plastic) that applies a lot of inward pressure to help avoid any air gaps. Ambienceaudio CT-1 Phone Cable operates under a similar principal- "Pinlok RCA connector - Traditional RCA connectors cause the signal to micro arc across the resistive pathway of the connection. Our patent pending PinLoK RCA connector has an oversized pin that will compress to enter a normal RCA socket. Spring tension then continues to push and expand the inserted pin for maximum contact pressure. This increase in surface contact lowers the amount of microarcing and reduces distortion". These designs take the locking WBT style plugs to the next level. As long as there is enough surface area of bonding/very, very small gaps. And for these lower level signals, the surface area of bonding/very, very small gaps is less than most people would think. In other words, the "small" area of bonding/very, very small gaps resulting from a small sphere or "pin" is plenty. Lets not forget in traditional RCAs (or whatever), the real area of contact is almost for certain <<2% of the apparently area of contact. With starting clean surfaces (and more than likely even without, though it wont be nearly as much, amount of contact pressure where asperities (peaks on mating surfaces) meet is key), there will be some bonding. Until it gets crudded up from stuff in air, which starts happening the second you connect. Hence the recommendation to clean and reapply enhancer every few months. Even just a wiggle/twist/remove/replace if you dont have enhancer or cleaner is a good idea. Scrub off some crud and make some new/different mountains meet and create fresh bonds. For a while at least.

“Initially two tribo-surfaces (disk and counterface in this research) are covered with layers of oxides, sulfides, and other solid compounds. On top of these compounds there are films of adsorbed gases and hydrocarbons. Without these surface layers, the mating surfaces could bond more strongly......
When two surfaces are mated to one another, actual contact only occurs at various isolated points (the asperities) called junctions. When the areas of all the junctions are summed, one obtains the real area of contact. On the other hand, the area of contact which one determines through geometrical considerations of the actual part on a macroscopic level is called the apparent area of contact. The real area of contact is usually much less than the apparent area -- sometimes a thousand or more times less [11]. Typically the real area is about 1% of the apparent area. The diameter of typical junctions has been estimated to range from 1 micrometer to 100 micrometers [12].
Considering only the real area of contact, it is the protruding asperities which will make contact with the opposing surface (and its protruding asperities). On an atomic level, because a force is being applied to push two surfaces together, bonding will occur somewhere in the junction if the two materials have compatible structures. The extent of cleanliness and other factors such as temperature and environment also play a role. This is due to the nature of atomic forces.
Figure 1 shows the force versus distance between two surfaces [13] which is similar to the inter-atomic force vs. distance between two atoms [14]. Being in the trough signifies bonding and somewhere in any given junction there will be atoms which are in the trough. Thus there will almost always be some bonding between two unlubricated surfaces when an external force is applied. How much bonding occurs over a given junction depends on many things including the force applied, the compatibility of the crystal structures, material properties such as Young's modulus, how closely the two surfaces mate to one another on an atomic level and how clean the surfaces are. As two surfaces are properly cleaned the friction coefficient can become very large due to adhesion effects, sometimes even exceeding 100 [15]. This type of cleaning can involve sputtering in a vacuum chamber.
If one considers two surfaces, each containing an infinite number of atoms being pressed together, statistics tells us that at some location in the interface of the two surfaces, the atoms of the opposing materials will be properly positioned and will lie in the trough of Figure 1. This creates a bond.”

With “digispeakers”, (except when fed analog sources), the above is moot. Though even with analog sources (for Meadowlark Speakers at least) the speaker cable interfaces (and cables themselves) are eliminated as they have analog (and digi of course) inputs on the speaker.

Let the cable guys discuss/debate all they want about interconnects and speaker cables. They will still never get a grade of “100%”. Speakers that have digi inputs, like the Meadowlarks, make analog cables obsolete (other than interconnects for TT, RtR, cassette, and other lesser used analog sources), thus getting a score of "100%". Most people listen to digi sources these days so this is huge.

2. The steep (typically 48 dB/octave) high pass DSP filters (and low pass though not nearly as important to final sound quality) to block out detrimental energy to each speaker in the speaker. This is an improvement over less steep conventional analog high/low pass filters. In addition, doing this in the digital domain in and of itself is a significant improvement than doing it analog domain (less heat, better longevity, etc). I think digital technology is very, very close to (basically we’re there and next the price is going to come down) as good as its gonna get with respect to how well the vast majority of people can hear.

Continued in post #67, I just edited and went over 10,000 characters.
 
Last edited:

Pat McGinty

Senior Member
Not to throw stones at other designers or designs, but I don't get the impression that you are crazy about large (say 10 inch or greater), high sensitivity woofers. Have you ever played with setups like that? Ever tried very high sensitivity woofers?
uh? One does not get high sensitivity without giving up something that one also wants. Allowing that we're working with equal flux densities, you buy sensitivity by shortening the coil so that a greater percentage of it is in the gap, resulting in more force per volt. Obviously, you're giving up displacement which, for me. is the single most valuable characteristic in a woofer.

Why would you make that trade? Amp power is dirt cheap, so what's the point?

You're gonna take up a bunch of space on the baffle to do work you could've done with a smaller driver. So the cabinet grows for no good reason, hence the cost. You've seen old fashioned designs like that, recently. no?

OK, so let's go your way, so far you've given up bass, increased cabinet size, ugliness and cost....and why? To save a couple of hundred bucks so you can avoid buying a real amp?

Pleeeeease! That's so 20th century.

Maybe you'd like a nice 90HP automobile withlittle, skinny bicycle tires to reduce rolling resistance? Give it up man, buy the amp power and be happy.
Not to throw stones at other designers or designs, but I don't get the impression that you are crazy about large (say 10 inch or greater), high sensitivity woofers. Have you ever played with setups like that? Ever tried very high sensitivity woofers?
Huh? One does not get high sensitivity without giving up something that one also wants. Allowing that we're working with equal flux densities, you buy sensitivity by shortening the coil so that a greater percentage of it is in the gap, resulting in more force per volt. Obviously, you're giving up displacement which, for me. is the single most valuable characteristic in a woofer.

Why would you make that trade? Amp power is dirt cheap, so what's the point?

You're gonna take up a bunch of space on the baffle to do work you could've done with a smaller driver. So the cabinet grows for no good reason, hence the cost. You've seen old fashioned designs like that, no?

OK, so let's go your way, so far you've given up bass, increased cabinet size, ugliness and cost....and why? To save a couple of hundred bucks so you can avoid buying a real amp?

Pleeeeease! That's so 20th century.

Maybe you'd like a nice 90HP automobile with little, skinny, bicycle tyres to reduce the rolling resistance?

Knock it off, get over it and buy enough power.
 

Prime Minister

Site Owner
Staff member
uh? One does not get high sensitivity without giving up something that one also wants. Allowing that we're working with equal flux densities, you buy sensitivity by shortening the coil so that a greater percentage of it is in the gap, resulting in more force per volt. Obviously, you're giving up displacement which, for me. is the single most valuable characteristic in a woofer.

Why would you make that trade? Amp power is dirt cheap, so what's the point?

You're gonna take up a bunch of space on the baffle to do work you could've done with a smaller driver. So the cabinet grows for no good reason, hence the cost. You've seen old fashioned designs like that, recently. no?

OK, so let's go your way, so far you've given up bass, increased cabinet size, ugliness and cost....and why? To save a couple of hundred bucks so you can avoid buying a real amp?

Pleeeeease! That's so 20th century.

Maybe you'd like a nice 90HP automobile withlittle, skinny bicycle tires to reduce rolling resistance? Give it up man, buy the amp power and be happy.


Huh? One does not get high sensitivity without giving up something that one also wants. Allowing that we're working with equal flux densities, you buy sensitivity by shortening the coil so that a greater percentage of it is in the gap, resulting in more force per volt. Obviously, you're giving up displacement which, for me. is the single most valuable characteristic in a woofer.

Why would you make that trade? Amp power is dirt cheap, so what's the point?

You're gonna take up a bunch of space on the baffle to do work you could've done with a smaller driver. So the cabinet grows for no good reason, hence the cost. You've seen old fashioned designs like that, no?

OK, so let's go your way, so far you've given up bass, increased cabinet size, ugliness and cost....and why? To save a couple of hundred bucks so you can avoid buying a real amp?

Pleeeeease! That's so 20th century.

Maybe you'd like a nice 90HP automobile with little, skinny, bicycle tyres to reduce the rolling resistance?

Knock it off, get over it and buy enough power.
But....
Or maybe, However....

There is one thing those big beasts do that I have never heard a speaker with a small driver do. Namely, playing low volumes, 60-70 db, convincingly and well. I'm not saying they can't, but I've yet to find one. They all seem to need some power to get moving and then find their groove. Those big, highly sensitive, drivers seem to be able to breathe the breath of life into music at those silly low levels. Ella is sitting right in your room, late at night, almost whispering But Not For Me into your ear. You can almost feel Frank's breath on the mic when he singing In The Wee Small Hours. Lots of speakers do that well when they are moving, but I've only heard those big beasts do it in small rooms and low volumes.
 

Pat McGinty

Senior Member
Sorry, I didn't mean to be so abrupt. Just me, the #1 priority is getting believable vocals at realistic levels. Again, just me, lower levels are of less interest because, to me, they sound distant or miniature. I guess it comes under the objective: to sound exactly like the real thing.

I think that the large driver two way is really a work-around for the problems that occur in three ways. Ideally, for midrange reproduction, you'd like a coercive motor, very light moving system and very soft suspension. You don't need excursion. You do need fine resolution around the center of rest. So you'd think that the problem would be easy to solve, but no. It's so darn messy that guys, somewhat sensibly, have pushed forward with the two way idea. And our perceptions do tell us that there's a "rightness" to the two way that often lacks in three ways. So, yeah, big woofer two way, I get it, like it.

So what's so tough about the three way? It's not in obtaining great transducers; there are plenty. The problem is the passive bandpass filter. For a for pole bandpass you'll - at a bare minimum - have this:

passive bandpass.jpg


Without dragging thru all of the details, this circuit is your defeat. And the parts, if you use good ones, are gonna cost more than the transducer. So, yeah, put that mess up against your much simpler two way and you're going to pick the two way.

Most guys have never heard a great midrange driver without a passive bandpass in the way. So, yeah, the preference for two ways is pretty well ingrained and quite understandable.

Back in the day, I worked long and hard to execute first order bandpasses. One C, one L one R. I'd like to think I pulled it off. But you can count on one hand the number of designers who managed to make that topology work. Richard Vandersteen, Jim Theil. That's about it, because it's unbelievably difficult.

Fortunately, and HALLELUJAH, we're moving past this stumbling block and into a world where three ways and four ways can exceed anything that can be done with a two way. That said, yes indeed, two ways can also be better than they've ever been, too.
 

Prime Minister

Site Owner
Staff member
Sorry, I didn't mean to be so abrupt. Just me, the #1 priority is getting believable vocals at realistic levels. Again, just me, lower levels are of less interest because, to me, they sound distant or miniature. I guess it comes under the objective: to sound exactly like the real thing.

I think that the large driver two way is really a work-around for the problems that occur in three ways. Ideally, for midrange reproduction, you'd like a coercive motor, very light moving system and very soft suspension. You don't need excursion. You do need fine resolution around the center of rest. So you'd think that the problem would be easy to solve, but no. It's so darn messy that guys, somewhat sensibly, have pushed forward with the two way idea. And our perceptions do tell us that there's a "rightness" to the two way that often lacks in three ways. So, yeah, big woofer two way, I get it, like it.

So what's so tough about the three way? It's not in obtaining great transducers; there are plenty. The problem is the passive bandpass filter. For a for pole bandpass you'll - at a bare minimum - have this:

View attachment 16303


Without dragging thru all of the details, this circuit is your defeat. And the parts, if you use good ones, are gonna cost more than the transducer. So, yeah, put that mess up against your much simpler two way and you're going to pick the two way.

Most guys have never heard a great midrange driver without a passive bandpass in the way. So, yeah, the preference for two ways is pretty well ingrained and quite understandable.

Back in the day, I worked long and hard to execute first order bandpasses. One C, one L one R. I'd like to think I pulled it off. But you can count on one hand the number of designers who managed to make that topology work. Richard Vandersteen, Jim Theil. That's about it, because it's unbelievably difficult.

Fortunately, and HALLELUJAH, we're moving past this stumbling block and into a world where three ways and four ways can exceed anything that can be done with a two way. That said, yes indeed, two ways can also be better than they've ever been, too.
No need to apologize. You weren't being abrupt. I was being stubborn. :)

I get the idea of vocals at realistic listening levels. However, what is the realistic listening level for a Sinatra vocal? They say that Sinatra was the first singer of the microphone age. Ethel Merman could be heard at the back row of the Roxy, with no amplification. Sinatra, however, was the first major artist to get what you could do with a microphone, and a lot of his recordings were near enough whispered into a mic. Karen Carpenter didn't sing much above speaking level. It's that stuff that it seems hard for small two ways to do well.

I listened to a near enough $20k system last week (speakers and amp only), that did this very well. A giant two way with a 10 inch woofer, a dome tweeter and about 15 watts of tube power. And Ella, at very low levels, was just in the room. And it still had great dynamics. And, it was the better part of $20k.

Now, if I understand you correctly, the modern approach to doing this would be a DSP 3 way, with a really killer midrange driver. But could it do this at a low level, glass of cognac in hand, late night, like 65 db average, listening session?
 

Pat McGinty

Senior Member
Yeah, the issue is the behavior of any ten incher above a kilohertz. Too much mass, expect cone flexure. To precisely deal with the mid band you want a low moving mass mid. This one's nice:
or this:

Running a mid without the passive circuit changes everything, especially the resolution because there's an absence of smearing and time domain perturbations. Just a nice, tight grip by the amp to the coil. You're looking for a loose suspension and high BL product to realize that advantage. Remember that we now have fine control over what happens in the pass band - so we can go right ahead and nail it.

Hey, if you're running software volume control into Kites, you'll want to adjust the Input level on your low level Preset so that your app's vol control is usually working in at the top 90-100%. As you know, software volume control works by bit dithering - so you want to keep that to a minimum.

Three volume three way. The lower compartment, open to the rear, is for hiding your streamer. Ready for final glue up.

20190911_111149.jpg
 
Top