[Duke]

Part One

My top-of-the-line speaker [as of June 2008], the Dream Maker, uses a configuration that to the best of my knowledge hasn't been used before.  It's more of an evolutionary than a revolutionary step, but I think it brings together a desirable and perhaps unique set of attributes.

Let me first describe "the goal", as I see it, and then describe how I tried to approach this goal.

Most of the difference between the concert hall and the listening room lies in what’s happening in the reverberant field.  In the concert hall, there’s a fairly long time delay between the first-arrival sound and the arrival of significant reverberant energy.  And then when that reverberant energy arrives, it is powerful and rich and diffuse, and its spectral balance is similar to that of the first-arrival sound (only modified by the room surfaces it has encountered), and finally it decays rather slowly.  It is this rich, powerful, diffuse, slowly-decaying reverberant field that gives music in a good venue that delicious aliveness and lushness that thrills and envelopes us.

In contrast, in a typical home listening room the first reflections begin to arrive after only a fairly short time delay, and they tend to be distinct rather than diffuse.  The relative amount of energy in the reverberant field is much less than in the concert hall, the spectral balance is usually wrong (being dominated by the speaker’s omnidirectional output, which is seldom smooth), and the reverberant field dies away fairly quickly. 

So, my primary goal was to put more energy out into the reverberant field, in a way that is beneficial rather than detrimental.

Attempting to more closely approximate the reverberant field of a live venue is nothing new - Amar Bose did it, but in my opinion he didn’t do it right; the 901 puts too much energy into the reverberant field, and it doesn’t use high-resolution drivers.  Omnidirectional and polydirectional systems (Richard Shahinian came up with the latter term) in my opinion do a better job with getting the correct balance between direct and reverberant energy.  Wide-pattern dipoles (like the SoundLab A-1) or even wide-pattern monopoles (Sonic Weld and Beveridge come to mind) also give a much richer than average reverberant field, and from my experience as a SoundLab dealer I think that what’s happening in the reverberant field plays a much larger role than is generally appreciated.

Now there are several formats that would put a lot of energy out into the reverberant field - dipole, bipole, and omni or polydirectional being the main ones.   Why did I decide to do a bipolar system? 

Well, a dipole either has to be huge or equalized to have good bass extension, and in either case the displacement requirement for good bass extension at high volume levels is pretty imposing. And if equalization is required, that limits amplifier choices - I wanted my speaker to work well with SET and OTL tube amps. 

One problem with omnidirectional systems are the room requirements.  You see, we’d like for the onset of reflections to be delayed by a good 10 milliseconds after the first-arrival sound, and this corresponds to a path length difference of eleven feet.  So, ideally, an omni should be placed far enough away from the walls to avoid early reflections arriving within that time interval (okay there’s the floor bounce and probably the ceiling bounce, but those are harder to get rid of).   Most of us don’t have rooms that big.  This same requirement holds true for dipoles - indeed, most Maggie and Quad and Martin Logan and SoundLab and Apogee owners find their speakers sound best when positioned a good five or six feet out from the wall behind the speakers (but placement close to the sidewalls is not a problem, because dipoles have a null to the side).   

Omnis also have a problem with off-centerline soundstaging.  While the soundstage holds up better than normal for off-centerline listeners, with some recordings they will give you “ten-foot-wide guitar and/or ten-foot-wide mouth” syndrome.   I find that distracting. 

So, what do bipoles do that’s so special?  And, perhaps more important, what are their drawbacks?  (Ha!  You don’t see that in ad copy every day, now do you??)

[Duke]

Part Two

A bipole isn’t really special unless you do it “right”.  And, as you probably have guessed, I have a definite opinion as to what constitutes a “bipole done right”.

Obvoiusly, a bipole has rear-firing drivers, which put out more energy into the reverberant field.  But unless that extra reverberant field energy is spectrally correct and late-arriving, it’s more likely to be detrimental than beneficial.  And unlike a dipole, the bass coming from the back of the speaker doesn’t cancel out the bass coming from the front of the speaker, which might be a good thing - but, in order to get that, you gotta have a box, and a box adds cost and the potential for (all you dipole fans know this one...) “boxiness”. 

So the first thing we have to do to a bipole is control the radiation pattern so that our reverberant field energy is spectrally correct.  I do this by using a low-coloration 90-degree constant-directivity waveguide for the top end, crossed over to a fairly large woofer at the frequency where the woofer’s pattern has narrowed to about 90 degrees.  True the woofer’s pattern widens some at low frequencies, but the ear is relatively forgiving of this.  I get good pattern control down to 1500 Hz; I’d like to go down to 500 Hz but that would take much larger woofers and waveguides which add more size and cost than I think is worthwhile.

Those of you somewhat familiar with the SoundLab A-1 may recall that it also has a radiation pattern that’s 90 degrees wide, both front and back (though it’s a line source, not a point source).  My use of a similar pattern is no accident - I have a great deal of respect for Roger West’s designs, and if you want to accuse me of trying to build a cheap imitation of the SoundLabs I’d take that as a compliment.

One beneficial by-product of a well-controlled 90-degree radiation pattern is that, when toed in by about 45 degrees, it gives remarkably good off-centerline soundstaging.  You see, the ear localizes sound by two mechanisms: arrival time and intensity.  With this extreme toe-in, for the off-centerline listener the near speaker wins arrival time but the far speaker wins intensity because the listener is now much more on-axis of that far speaker.  This will not work well with most conventional speakers - their radiation patterns vary too much up and down the spectrum.  Also, this format prevents ten-foot-wide guitar and ten-foot-wide mouth syndrome.   The off-axis soundstaging isn’t perfect, but it’s much better than with conventional speakers - and the tonal balance is correct from anywhere in the room.

Another benefit of this well-controlled radiation pattern is that we are much less likely to need room treatment, assuming we don’t have a severe problem like slap-echo.  Just as a piano doesn’t need room treatment to sound like a real piano, so too a loudspeaker shouldn’t need room treatment to recreate the sound of that piano. I don’t use room treatment at audio shows, and my rooms seem to be pretty popular (okay, I confess -  I use middle-aged men as bass traps!).  This is not to say that room acoustics wouldn’t benefit from appropriate attention, but with controlled-pattern speakers diffusion panels would usually be a much better choice than absorption panels.

The next thing we must do to have a good bipole is keep the box from sounding like a box, or else our friends with dipoles will laugh in our general direction.  So we need a solid, dead box with good suppression of internal reflections.  Well that’s not rocket science, but neither does it lead to lightweight, inexpensive speakers.  Next we need to avoid diffraction, as this will add a boxy signature even to a dipole (yup, I once auditioned a $19k dipole speaker that sounded boxy).  We do get one anti-boxiness freebie: bi-directional sound propagation, in and of itself, inherently tends to sound less like it’s coming from a box. 

[Duke]

Part Three

Finally, there is problem that arises in the lower midrange region with a bipole.  We need to understand and deal with this problem (or at least I do, and you can pretend like you’re playing along).  Here’s what happens: As the wavelengths become long enough that the rear-facing woofer no longer has good directional control, the sound starts to wrap around the enclosure and interact with the output from the front-facing woofer.  At the frequency where the wrap-around path length is equal to one-half wavelength, the output from the rear woofer arrives 180 degrees out-of-phase with the front woofer, and a cancellation notch results.  This notch grows shallower as you move farther off-axis, and it disappears in the power response, but it’s there in the first-arrival sound.. 

Different manufacturers of bipolar loudspeakers have dealt with the wrap-around notch in different and often innovative ways.  Mirage used a cabinet that was much wider than it was deep, and I think they also notched the response of the rear-facing woofer so that it wouldn’t cancel the front woofer at the wraparound frequency.  Definitive Technology patented using a single side-firing woofer with bipolar midrange and tweeter drivers, and this allowed them to go with a much  narrower cabinet than the Mirage approach.  In their “5" series loudspeakers, Genesis has a rear-facing midbass driver that is out of phase with the front-facing one, so that the wrap-around energy is reinforcing rather than cancelling the output of the front woofer in that lower midrange region.  Louis Chochos of Omega uses a cabinet that again is wider than it is deep, as the wide cabinet face reduces the amount of wrap-around energy at any given frequency while the shallowness of the cabinet raises the frequency where the wrap-around arrives out-of-phase, and the higher that frequency the better the wide cabinet is at pattern control.  The result of the Omega approach is the dip isn’t very deep, but it’s still there.  In practice, it’s not audible because dips are harder to hear than peaks and the reverberant energy fills it in anyway, which psychoacoustically works in that frequency region. 

My approach is the same as what the Omegas use - my box is quite a bit wider than it is deep.  Actually I came up with this format back in the 90's, but didn’t know about waveguides for pattern control so I didn’t have enough pieces of the puzzle to build my dream speaker back then. 

Okay if the wraparound energy from the rear-facing woofer is such a pain in the output stage, why not just do a dipole?  Well, that wrap-around energy does some good things as well. 

First of all, it largely eliminates the “baffle step”, once the wrap-around notch has been addressed.  Baffle step compensation is beneficial if you listen up close to conventional speakers (where the first-arrival sound dominates the perceived tonal balance), but detrimental if you listen from farther away (where the now-bass-boosted power response dominates).  With a good bipole, you can have it both ways with no downside.  So instruments with a lot of midbass energy, like cello and guitar, have nice “body” and natural-sounding fullness.  Second, if done right (and don’t you know it - I’m gonna “do it right”), the wrap-around energy can actually reduce the magnitude of the floor-bounce notch.  The way we do this is by having the rear-firing woofer at a different height than the front-firing woofer.  Actually, this is a very important - and I think unique - aspect of my particular variation on the bipolar theme, which we’ll come back to in the next paragraph.  The third benefit of the wraparound is that it reinforces the bass somewhat, which is nice because a bipole usually gets less boundary reinforcement than a conventional speaker because it wants to be well out in the room away from the wall that’s behind it.

One more benefit accrues from the rear woofer being offset to a significantly different height than the front woofer: Smoother bass.  Have you ever noticed that, particularly with a subwoofer, moving the bass source just a little bit can have a much larger-than-expected effect on the perceived bass response?  Well, the reason is this: A low frequency source at a given location will interact with the room in a way unique to that location to produce a unique room-induced peak-and-dip pattern at the listening position.  The more low frequency sources spread around the room, the more they average out to a nice, smooth bass response at the listening position (this is the basis for my Swarm four-piece subwoofer system).  When the Dream Makers are toed in as recommended, the two woofers on each speaker are displaced in all three dimensions relative to one another.  While this isn’t as effective at smoothing the bass as spreading them far apart would be, it’s still a worthwhile improvement over just one woofer, or having two woofers on the front baffle.  When I was at Robert Greene’s house to pick up the Dream Makers, he told me that he was impressed by the smoothness of their measured in-room response, particularly in the bass region.  He said they were competitive with some of the best he’s seen in this respect, and that would include the Gradient 1.3 and Revolution. 

[Duke]

Part Four

So, let us recap.  When positioned properly, the controlled-pattern offset bipole gives a well-energized, spectrally correct, late-arriving reverberant field without adding detrimental early reflections, and does so in a smaller room than an omni can.  The midbass region is well supported, the bass region is smoother than what most speakers can do, the sweet spot is extremely wide, and the tonal balance is correct throughout the room, all without room treatment in most cases.   No equalization is required, dynamic contrast is preserved, and low-powered specialty tube amps can be used.

In the Dream Makers, I tried hard to make a speaker that works WITH the room, and also WITH the way the ear/brain system processes sound, as much as possible. My hope was to come up with an acoustically and psychoacoustically well thought-out concept, and then execute it well.

Now for those of you who are saying “sure that’s nice, but no way am I gonna drop nine grand on a pair of speakers”, many of the best attributes of the Dream Makers can be realized in a smaller and less expensive system.  In fact, I’ve got one in the works now.

I welcome questions and comments.  And if you actually read this far... dude, get a life!

Duke

[rockadanny]

And if you actually read this far... dude, get a life!

Doh! That's what people keep telling me! Great info and extremely well written. Thanks Duke!
Now get off'n that computer and get woikin' on that more affordable version ... now!!

[Duke]

Thanks, Danny!

You're definitely hard-core. 

Eventually I think there will be two "more affordable" variations of the Dream Maker - one around 2.5 grand, and one around 5 grand.  I'm doing the 2.5 grand one first.  Still have some optimizing to do, but if all goes well we're maybe two months away from a production pair. [<- that proved to be comically optimistic]

Duke

[gsm18439]

Great information and very well-written. However, in this discussion you did not address the issue of speaker placement within the room - especially with regards to front and side walls, short vs long-wall placement, and near-field vs far-field listening (or, should I say, ideal distances from the speakers to the listening position(s)). Thanks.

[Canyoneagle]

GREAT stuff, Duke.
I look forward to hearing a pair!
Michael

[Duke]

Gsm18439, good questions. 

With any bipole or dipole, normally I recommend a good five feet or so between the back of the speaker and the wall (3.5 feet is an approximate minimum based on experience).  The reason is, the ear tends to classify reflections arriving earlier than 10 milliseconds after the direct sound as coloration, and such early reflections can also mess up the soundstaging.  After 10 milliseconds, the reflected energy is generally beneficial - assuming its spectral balance is in the ballpark.  And note that 10 milliseconds isn't a hard barrier; it's in a fuzzy transition zone. 

Sidewall placement is much less critical, assuming adequate distance between speaker and the wall behind it.  I design for about 45 degrees of toe-in, which pretty much avoids a strong early sidewall reflection.  There will be a later-arriving sidewall reflection off the far sidewall, but that's likely to be beneficial from a spaciousness standpoint because it's arriving at the opposite ear from where that signal was first heard. 

I haven't had them in a room where I could try longwall placement, and see if the absence of nearby sidewalls would let me get away with much closer placement to the wall behind the speakers.  But I tried that with SoundLabs once and it worked pretty well.  I think you'd want to avoid backing them into corners.

Minimum listening distance is related more to the vertical spacing of the drivers and the crossover frequency.  I think that 6-7 feet is probably the minimum for good integration.  If the crossover were lower (as in SP Tech speakers), the minimum listening distance would be less because the ear is not very good at resolving the height of a sound source below 1000 Hz, and it gets progressively more forgiving in this respect as we go down from there.

I would say the ideal listening distance is probably at and/or a few feet behind the equilateral triangle point (with 45 degrees of toe-in, the speakers' axes will be crossing in front of the listening position).  But the tonal balance will hold up quite well throughout the room, and even into the next room.

Duke

[Russell Dawkins]

Robert E Greene of The Absolute Sound just an hour ago posted a flattering comment on this design in reply to the question "why did this speaker get a Golden Year Award?"
http://groups.yahoo.com/group/regsaudioforum/message/23243

[Duke]

Thank you very much for noticing that and posting the link, Russell! 

Duke

[JohnR]

I only just read this now  Very nice, clear explanation, Duke! I always just assumed bipoles were trying to approximate a (full spectrum) monopole. How wrong I was

[ttan98]

Duke,

After reading through your 4 parts tutorial on bipole design, you did not mention any guideline on how to determine the offset. I assume the offset is vertical only no horizontal offset. Is there a general guide on how to determine the offset?

I have drivers ready to build a bipole, a 2-way speaker system hence 4 drivers pers side. the tweeter is dome type and woofer a std. 5 1/4" driver.
I believe the rear drivers, the tweeter and woofer should be mounted adjacent to each other and displaced/offset vertically down in respect to the front drivers. How much offset? My x-over freq is about 2Khz.

Any advice is welcomed before I start cutting.

Thanks.   

[Duke]

At some point I may write up something more polished and formal; the above is more of an infomercial-style sales pitch.  I've been fairly secretive about exactly what's happening on the back of my bipolars, even to the point of not allowing people to photograph the back at audio shows.  But I do let them take a look.

If you were to take a look, you'd see that both of the drivers on the rear of the speaker are shifted down closer to the floor.  That means nothing to most of the people who've looked behind my speakers, but I'm sure it means something to you.  In fact that's exactly what you were already planning to do, isn't it?  Nice job!!

By placing the rear woofer down low it gets pretty good boundary reinforcement in the bass region.  There's another benefit as well, and that is a partial filling-in of the floor bounce notch.  The front woofer still floor-bounce-notches, but the rear woofer doesn't so the net result is a shallow dip instead of a deep notch.

To answer your question, I suggest making your best estimate as to how much boundary reinforcement you want the rear woofer to get.  The closer to the floor you place the rear woofer, the more bass you get (I'm sure you long since figured that out, but just wanted to double-check in case there was some unforseen issue).  I put my input terminal cup below the rear woofer, so that puts that woofer a few inches above the floor. 

Very best of luck with your project!

Duke

[Skip Pack]

Duke,

I remember hearing the Dream Makers at the 2008 LSAF. My wife and
I enjoyed them very much. We (me particularly) also enjoyed the Altec 605's
that John Busch had in open baffles with 18" woofers. Since I figured I could
probably buy and have rebuilt some Altecs for an amount I could afford, whereas
I cannot imagine touching your (very fair) price on the Dream Makers, I decided
to pursue the Altecs.

Reading your exposition in this posting, and Jim Griffin's exploration of your
concept leads me to consider an over-the-top idea.

The altec 604's and 605's seem to be frequently used in BR enclosures of 8-11
cu.ft. I'm wondering if this could work with your offset bipole concept. If you had
two volumes, about 28" high internally isolated horizontally with the front-firing
driver in the upper volume offset toward the bottom, and the rear-firing driver in the lower
volume, near the floor, it might work. I would expect the front-to-rear internal
dimension to be about 13", which would accommodate the deeper 604. The front
and back panels would be around 20" wide. The added volume would have to come
from the sides. The sides would be symmetrical front and rear and would angle back/forward
in two steps, perhaps a 10 degree angle for each step. The front and back angled
panels would end at an sade panel perhaps 4-8" wide that would be perpendicular to
the front and back baffle centers.

Thus, you end up with what amounts to two boxes, 50-60" wide, 28" high, and
13" deep in the center. They would be tapered center-to-side edge with the
angled panels, providing a long wraparound path and somewhat controlled
diffraction. The two would "stack" to about 5'.

I won't ask if this is silly, that's apparent. But, could it work well? I am constructing
a room where this could work, though I'm sure there aren't that many situations where it
would. Any thoughts are most welcome, and I will fully understand if you feel that
your comments might be too commercially revealing. In any case, thank you for your
interaction with our low-yield DIY community.

Skip Pack

[Duke]

Skip, I think the system you described would work well....  but I'm not gonna help you move it! 

I'd suggest using some sort of variable tuning system, as you might end up with a bit more low bass energy than the raw numbers suggest.  You see, the rear drivers' output combines in random phase over most of the spectrum, for a theoretical +3 dB compared to the output from the front-facing drivers alone, but in the low bass region when the wrap-around distance is less than 1/4 wavelength you start to get the rear driver's energy adding in-phase, so it can theoretically approach +6 dB in the low bass.  In practice I don't think you can count on an extra 3 dB in the bass region, but maybe an extra 1.5 dB or so.  The tradeoff is, of course, that the effective midband efficiency is correspondingly reduced relative to what it would be with all the drivers on the front, facing the listening position.   This effect disappears in the farfield where the reverberant sound dominates, but with an offset bipole there is less tonal discprepancy between the nearfield and farfield response because that "extra" wrap-around bass is offsetting the baffle-step, which disappears in the farfield.

[Marcus T]

Part One

My top-of-the-line speaker [as of June 2008], the Dream Maker, uses a configuration that to the best of my knowledge hasn't been used before.  It's more of an evolutionary than a revolutionary step, but I think it brings together a desirable and perhaps unique set of attributes.

Let me first describe "the goal", as I see it, and then describe how I tried to approach this goal.

Most of the difference between the concert hall and the listening room lies in what’s happening in the reverberant field.  In the concert hall, there’s a fairly long time delay between the first-arrival sound and the arrival of significant reverberant energy.  And then when that reverberant energy arrives, it is powerful and rich and diffuse, and its spectral balance is similar to that of the first-arrival sound (only modified by the room surfaces it has encountered), and finally it decays rather slowly.  It is this rich, powerful, diffuse, slowly-decaying reverberant field that gives music in a good venue that delicious aliveness and lushness that thrills and envelopes us.

In contrast, in a typical home listening room the first reflections begin to arrive after only a fairly short time delay, and they tend to be distinct rather than diffuse.  The relative amount of energy in the reverberant field is much less than in the concert hall, the spectral balance is usually wrong (being dominated by the speaker’s omnidirectional output, which is seldom smooth), and the reverberant field dies away fairly quickly. 

So, my primary goal was to put more energy out into the reverberant field, in a way that is beneficial rather than detrimental.

Attempting to more closely approximate the reverberant field of a live venue is nothing new - Amar Bose did it, but in my opinion he didn’t do it right; the 901 puts too much energy into the reverberant field, and it doesn’t use high-resolution drivers.  Omnidirectional and polydirectional systems (Richard Shahinian came up with the latter term) in my opinion do a better job with getting the correct balance between direct and reverberant energy.  Wide-pattern dipoles (like the SoundLab A-1) or even wide-pattern monopoles (Sonic Weld and Beveridge come to mind) also give a much richer than average reverberant field, and from my experience as a SoundLab dealer I think that what’s happening in the reverberant field plays a much larger role than is generally appreciated.

Now there are several formats that would put a lot of energy out into the reverberant field - dipole, bipole, and omni or polydirectional being the main ones.   Why did I decide to do a bipolar system? 

Well, a dipole either has to be huge or equalized to have good bass extension, and in either case the displacement requirement for good bass extension at high volume levels is pretty imposing. And if equalization is required, that limits amplifier choices - I wanted my speaker to work well with SET and OTL tube amps. 

One problem with omnidirectional systems are the room requirements.  You see, we’d like for the onset of reflections to be delayed by a good 10 milliseconds after the first-arrival sound, and this corresponds to a path length difference of eleven feet.  So, ideally, an omni should be placed far enough away from the walls to avoid early reflections arriving within that time interval (okay there’s the floor bounce and probably the ceiling bounce, but those are harder to get rid of).   Most of us don’t have rooms that big.  This same requirement holds true for dipoles - indeed, most Maggie and Quad and Martin Logan and SoundLab and Apogee owners find their speakers sound best when positioned a good five or six feet out from the wall behind the speakers (but placement close to the sidewalls is not a problem, because dipoles have a null to the side).   

Omnis also have a problem with off-centerline soundstaging.  While the soundstage holds up better than normal for off-centerline listeners, with some recordings they will give you “ten-foot-wide guitar and/or ten-foot-wide mouth” syndrome.   I find that distracting. 

So, what do bipoles do that’s so special?  And, perhaps more important, what are their drawbacks?  (Ha!  You don’t see that in ad copy every day, now do you??)

Hello Duke etc,

Absolute love bipols / dipols and wish to build a pair using a couple of SEAS FA22RCZs that I have in each cab. Would really appreciate if anyone has any cabinet plans I could use for these speakers and where the rear speaker should sit, bottom or at the top. Think at the top would give a more spacial feeling ?

Thanks

[Duke]

Absolute love bipols / dipols and wish to build a pair using a couple of SEAS FA22RCZs that I have in each cab. Would really appreciate if anyone has any cabinet plans I could use for these speakers and where the rear speaker should sit, bottom or at the top. Think at the top would give a more spacial feeling ?

You can place the rear driver high on the back, but you'd lose the advantage of being able to fill in the floor-bounce dip because it will be floor-bounce dipping in the same general region as the front-facing driver.   See the response curves posted in my article:

http://www.hifizine.com/2010/06/the-controlled-pattern-offset-bipole-loudspeaker/

As for enclosure size, take into account the fact that you'll be getting in-phase boost below the wrap-around frequency, so size and tune the box accordingly.  Again, see the article for more information.

Best of luck with your project!