Designing a Reference Speakers - 4 Way Active

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I am starting to design a reference grade loudspeaker. In the following text I will be describing OB, TL, ESL and conventional box designs, 3-way or 4-way, active or passive XOs, etc from a new DIYer’s perspective. I wish you speaker building experts can point out my errors, contribute your ideas and make a project like this possible. It won’t just benefit me. I am sure it will also benefit many other new DIYers as well.

I have been interested in audio for 10 years. I have spent around US$30,000 on mid grade commercial audiophile equipment (first & second hand, including HT) and changed a number of systems. Since all of my power amplifiers were blown-up recently, and my last purchased speakers was found not really of my taste, I am now starting from scratch again. I am possibly one of a few unfortunate audiophiles that have spent that much money while still ended up with almost nothing. With my new family responsibilities I am no longer in the financial position to purchase commercial products any more - I mean commercial products that satisfy my “spoiled” ears. I know, of course, the MBL sounds fantastic, the Audio Note SE DHT, the Halcro or the Audio Research VT300 all sound fantastic. But unless I can win Lotto or have a money printing machine, there is no way I can afford those items. Fortunately, I have just discovered DIY audio – How lucky I could have been if I could have got into DIY audio 10 years earlier! For now, I believe that I will eventually achieve my goals with DIY and will have a first class system to enjoy for the many years to come. I believe that DIY will not only save me money but will also give me tremendous pleasure and in the future proud achievements. I certainly hope that will be the case.

I have spent 3 months searching on the Internet and have read many wonderful threads on this site. I would like to present my thoughts while I am thinking about how to choose / build my reference speakers. You experienced DIYers please correct my mistakes and point me to the right direction.

I am planning to build a new reference grade stereo system. The first thing on my list is a pair of speakers. My budget is up to US$5,000 for speaker drivers.

(I have been living in Australia and the UK for the past 12 years. However, English is not my native Language and I started learning English when I was 28 so please excuse me if you find errors in my grammar.)


Sound Characteristics:

I acknowledge that no speakers sound perfect and all reference grade speakers sound different. They have their own strengths and weaknesses. Also, what is good sound and bad sound is indeed highly subjective. I observed that there are broadly 2 types of speaker sound:

1. Some older audiophile speakers can sound very warm and sweet. They may not have supersonic high frequencies or subsonic low frequencies. The sound is often polite and inoffensive. Female vocals or violins can sound very beautiful, possible due to the nice colorations or distortions. You can sit for hours and hours in front of the music and never feel tired. However, the bass can be slow. They may not have enough speed and dynamics and may not give you the real thrill and impact for some types of music.

2. Some modern speakers, probably due to the use of modern advance drivers, can reveal a lot more details and dynamics in the music. Speed is much faster. Accuracy may be higher. The highs go high and the lows go low. However, comparing with the older style speakers, they can sometimes sound a bit harsh and lack a certain type of coherence, smoothness and warmth in the music.


Requirements

Considering the above, I have set the project requirements to be the following:

I would like my speakers to be full range covering from 20Hz to 50KHz with reasonably flat frequency response. They are designed for (with some compromises) all types of music. If I have to compromise (I know I have to), I prefer them to sound best for music in this order: (1) classical music including chamber music, violins, pianos, operas, powerful symphonies, organ music; (2) jazz; (3) vocals; (4) rock; and (5) ambient music. My current CD collections are 50% classical, 30% Jazz, 20% others. The speakers should be neutral sounding with accuracy, speed and “life-like” sound as well as slight warmth.

My room is about 7 meters by 9 meters.


Speaker Types:

1. ESL – I think it is much too hard for a new DIYer to build a pair of ESL. In any case, except the latest full range ESLs made by some of the most prestige names, most ESLs are brilliant in only the mid range frequencies. Even if I could build a pair of ESL, I want to stay away from it because of ESL’s impedance curves - I may not be able to build or afford to buy a very powerful SE DHT that can drive them well.

2. TL – I have been seriously thinking about it. It seems that the Ariel designed by Lynn Olson is one of the most popular TL projects. I have not had an opportunity to listen to it, but from what I have read, I believe the Ariel must sound really good. However, Olson openly confessed that the Ariel is not for everyone. He described the sound characteristic of the Ariel being similar to that of the early BBC monitors, of which I have a pair. The Ariel is not full range speakers so I don’t think it satisfies the requirement of this project. There are other TL DIY speakers out there, but I think unless I have listened to them I would not build them. This is partially because TL speakers are more difficult to build and far more difficult to tune. Another worry I have in mind is that although TL is renowned for producing low bass without using low bass drivers, in order for producing really true and deep bass a TL may need an enormously large enclosure. Being a DIY newbie I will have a greater risk of failure unless I use an existing, fully proven and well documented design like the Ariel.

3. Open Baffle / Dipole – I have to confess I am interested in a dipole more than anything else. The most popular one seems to be the ORION by SL. Again, I have not had an opportunity to listen to it. It is said that it is one of the best sounding speakers ever made. I like the concepts a lot, especially when considering dipole speakers having little dependence on room modes, its wide angle of sound dispersion and its “unboxed” sound. However, I am not sure if the esoteric tweeters and the midrange drivers used in the ORION are of my favorites. They are XOed at 1440Hz which means the tweeter has to do a lot of work for the upper midrange. Even though that particular tweeter is capable of crossing at such a low frequency according to its specs but I really don’t know how it sounds in that case. Don’t get me wrong. I am not criticizing the design. I am simply saying I don’t know the answer. I also don’t know if the two 10’ XSLs used in a dipole can produce enough SPL on the bass range between 40Hz to 120Hz or not. The objective measurement of the ORION may show flat frequency response, but the subjective listening experience may be different. By adding the Thor subwoofers with a single 12’ XSL will produce lower frequencies below 40Hz but it gives no enhancement to the 40-120Hz range. I am 100% sure that SL is a superior and very competent designer. My only worry is that when SL designed the ORION he might have taken costs into considerations hence made compromises on the design. For example, he recommends using relatively cheap amps to drive the ORION (yes, an active system has less demands on amps), and he had not chosen better drivers than the XSLs for the low frequency range. Therefore, I am only 99% but still not 100% confident about choosing the ORION to be my project. I know that something like the ORION can not be changed / modified easily unless you are very competent in redesigning the active crossover network and the EQ circuits. I am not competent enough yet so I won’t take the risks. Other OB speakers, like Steve Dodds’ BOB, a derivative from the ORION / Phoenix may sound better than the ORION. But for the above reason, I don’t think I am in the position to give it a go yet.

4. Conventional Boxed – the above make me to reconsider a conventional box design that I originally thought a bad idea. I have recently read comments that very well designed boxed speakers sound as good as any OBs or TLs.

So now I am back to thinking about a boxed design. Of course, if I can’t come up with a convincing design, I may go back to something like the ORION.


1 Way? 2 Ways? 3 Ways? Or 4 Ways?

I have briefly checked out the Single Driver site. It appears that using full or wide range drivers will be unable to produce powerful full range sound. If you use a full range driver and add a super tweeter, it will effectively become a 2-way. And if the bass is not enough, you may add a subwoofer, and you will effectively end up with a 3-way. Therefore, single driver is not the solution for my project.

Obviously it is hard for a 2-way to do full range. The popular project 2-way ProAc 2.5 clone may be a good project for somebody who likes rock music. However, it comes short if we consider our full range requirements.

Can a 3-way do the job? It surely can. However, if we use active XO, I believe that 4 way will be far superior. Here is why:

I would first choose the mid range driver to cover the most sensitive frequency range to the ears to avoid XO within the range. This means the mid driver will cover between 500Hz to 4,000Hz. I don’t want to use the mid driver to cover a wider range than that so that the mid driver can have an easier job. Note that it is within 1 decade / 3 octaves. A 4’ or 5’ driver will do the job superbly, producing very clean and sweet vocals and violins. I should have no problem of finding a high quality driver to fit in that role.

Obviously, many tweeters will be very happy to be XOed at 4,000Hz and take care of the 4,000-50,000Hz range.


(...see next post...)
 
Moderator,
Please put this post into my previous new thread. Thanks.

If I choose a 3-way, the base driver must cover between 20-500Hz. By using a box with the right volume, one driver can probably do that reasonably well. However, I suspect that the lower mid range will suffer a lot if the music contains a lot of notes below 80Hz. If I use 4-way, I would use a 24db LR XO at 80Hz, and use a 10’ driver to cover the 80-500Hz range, and 1 or 2 12’ drivers to cover the 20-80Hz range. This will help produce clean low-mid / upper-bass, and produce clean the powerful sub-base at the same time – the ultimate.

So I am heading to building a 4-way. Here is a summary:



XO: Passive or Active?

Of course, it has to be active.

One of the most difficult tasks in loudspeaker design is the crossover network. If I were to build a 2-way and if I could find 2 drivers that can XO very well, indeed I would probably consider using the simplest form of passive XO. But if I have chosen a 4-way, there is no way I would consider passive XO. It is nearly impossible. There are too many benefits of the active XO approach. They can be found in numerous sources thus are not listed here.

For a full range system, an active system is often in a different league and can provide far more superior sound comparing to a passive one. (I had both). As for costs, we DIYers can make a reasonably good power amplifier for about US$800. If we need 4 amplifiers to drive the speakers, we may need to spend $3,200 in total on amplifiers. That is still a little when considering the sound we can get. I bet that if you spend 5 times of that to get a Krell or Marklevenson to drive a passive speaker of 5 times of the price, you would be lucky to get half of sound from your 5 DIY amps with an active speaker.

The good thing is that there is no complication in an active XO design - People have already done it and we only need to use it. There are many sources from which we can order the industry standard, LR phase-linear 12db/24db XO PCBs, or even order a kit. Most kits use high quality Opamps and I believe (hope) that the sound quality is good, considering we do it electronically at the line level using active components instead of at the speaker using passive components.

I would consider 24db XO at 80Hz and 500Hz, and 24db or 12db at 4,000Hz.


Drivers

I would like to concentrate on a design approach for a new DIYer and don’t want to be distracted by driver discussions. If sufficient interests are generated in this project, I will start a new thread to discuss what drivers are the best for this design. I believe that since it is a 4-way design, it is relatively easy to find the right drivers.


Configuration

What configuration should I use? I would simply use:

Higher Mid
Tweeter
Lower Mid
Woofer.

I would initially use 4 separate boxes so that I may be able to do some experiments on physical time alignment at a later stage. The finished product may be enclosed in one box, with 4 separate internal chambers. The top 3 drivers will be placed as close to each other as possible. I will try to maintain the golden ratio 1:1.68 as much as practicable. They may be off-center. The woofer may be placed as close to the floor as possible.

I have read some comments on MTM configuration. It seems every design has its pros and cons. So I would stick with the simplest approach. I also noted that SL initially designed the Phoenix using MTMW, and changed it to TMW when he later designed the ORION.

I have a pair of WWMTMWW 7 unit tower active speakers. With certain types of music, they sounded as good (or as bad) as the JM Lab Grand Utopia, which, in my opinion, is a very ordinary speaker with a price tag of 6 digits while the MBL being cheaper is substantially better. My speakers sounded quite messy. When my first amplifier was blown up, I disconnected it from the top 2 woofers. Much to my surprise, the sound image was much improved. I realized that the tower speaker was too tall at 1.8 meters. My ceiling is about 3 meters high. The sound reflected from the ceiling to mess up everything. I concluded that for an ordinary home, having a very tall speaker is not a very good idea, unless you want to lay carpets on your ceiling.

Therefore, I think I will try to make my speakers shorter by not using any MTM type of configuration. I think for classical music, a TMW should sound better than a WMTMW anyway.


Box Types

I have read that sealed boxes are better sounding than base reflex. I have not heard any opposing opinions. I don’t make my speakers to sell but to own, therefore, size is not critical in my 7 x 9 meters room. Moreover, sealed boxes are far easy to make and tune.

So I would choose a sealed box.


Box Volumes

First, a tweeter doesn’t need a box because it is already boxed.

For the mid drivers, I would just use boxes that are large enough for their lower XO frequencies. I think that boxes are merely acoustically high pass filters. Because I will use active, very steep 24db XO, the signals feeding to the drivers will be within their designed FRs. Therefore, it can not go wrong with an oversized box for the mid drivers, even if some space may be wasted. In other words, the sizes of the boxes do not matter as long as they are large enough. It should not degrade the sound.

For the base drivers, I will consider a Q factor of about 0.7. In order to fit all the drivers into one short tower (with the tweeter at about 1.2m height?) and reduce the overall size – remember I use a sealed box, I may choose 2 12’ drivers in a basoteric configuration as in the Loudspeaker Design Cookbook to effectively reduce the size to half while at the same time improving the power output and reducing distortions.

Therefore, the box for the woofer does not need to be very large. I may initially make it smaller than suggested, then applying damping materials / fillings if I need to change the Q.


Phase Alignment

I believe that the active, LR phase-linear XO will take care of the issues. Because I use active XO, I have avoided all the phase problems associated most passive XOs.


Others
I am not so sure about time alignment. Some people think it is not important at all and it is only for marketing purposes. I think I will initially build separate boxes so that I can physically align them and see if it will make any difference.

As for diffraction, I am not sure about the actual impacts. Some people say it is very important. Some people doubt it. I guess that I may try to round up the box edge / corner, if possible.

I fully understand that the room acoustics is probably the most important factor contributing to sound quality (that is why I said earlier that I am interested in a dipole more than anything else). Nevertheless, things we can do about it are limited. So let’s not worry about room acoustics for this discussion.


Summary

1. I am planning to build a reference grade full range speaker.
2. Because I have not found an existing, proven design of OB, TL and ESL that fully satisfy the project requirements, I am exploring the option of a boxed design.
3. It will be a 4-way speaker with crossover points of 80Hz, 500Hz and 4,000Hz.
4. It will be an active system. The XO will be the standard LR 4th order phase-linear XOs.
5. The configuration is Higher Mid, Tweeter, Lower Mid and Woofer. Each is housed in separate box / chamber.
6. Oversized boxes / chambers are used for the mids. Q 0.7 is used for the woofer.
7. A Basoteric arrangement may be used for the woofer.
8. Phase alignment is not necessary because of the active XO. I may experiment time alignment. I may make round corners to reduce the effects of diffraction.


4 Reasons for Success

1. Traditionally, one of the most difficult tasks in loudspeaker design is the box design. By using an active 4th order XO and separate all drivers in different chambers, it has become a simple task.

2. Another very difficult task for traditional speaker design is the design of the XO. By using a standard active LR 4th order XO, I have avoided the complexities of maintaining phase, sensitivity, and a whole lot of other complicated issues. I have also avoided changing different caps, inductors and so on in order to find the perfect components that sound the best. They are usually very expensive.

3. Driver selection could have been a difficult task. By using an active 4th order XO in a 4-way design, each driver only needs to take care of a narrow bandwidth. It means that we don’t use the extreme ends of the driver response curves. The frequency response and impedance will be reasonably flat between XO points. Off-axis behaviors will be slightly less of a problem. We don’t need to worry about the problems associated with resonance frequency for the mids. It is relatively easy to find drivers that do the job well.

4. Driver and amplifier matching could have been critical. With 4th order XO in a 4-way, an amplifier will only need to work with a narrow bandwidth hence will perform substantially better, especially when the drivers have flatter impedance response. Therefore, amplifier selection is less critical. I could use a valve or a 40W class-A BJT amp to drive the tweeter, a sweet sounding 60W amplifier to drive the higher mid driver, a 150W class A/B transistor amp for the lower mid, and a 300W class A/B MOSFET amp to drive the woofers.


8 Reasons for Failure

Please provide your list here. Your critical comments will be very much appreciated.


The Challenge

Will this project be a success? Loudspeaker design is so complicated and fascinating. Are my views too simplistic? Can I simply use an active XO to avoid many complications in loudspeaker design? There are thousands of speakers out there, with many sounding ordinary. Will this speaker sound any different? Will it be just one of those? Will it be a reference speaker?

Many Thanks.
 
Tough & massive specs indeed

:eek: :)

I also listen to classical and covet the sound of a real full-range speaker system (who doesn't after all). And yes, such as they are, commercial offerings are way too expensive. So, diy-ing a worl-class speaker was the way to go for me too... until I set down to actually do it.

I don't want to seem reductionist -- but here are a few ideas drawn from my experience and reading yr posts:

(It goes w/out saying that you'll need access to testing equipment and s/ware, and a good idea of various drive-units by a variety of manufacturers...)

Targeting a SOTA, world-class speaker is great, but there's fun and good sound to be had improving a good kit -- that will be the base for an even better speaker, & so on. This site is an example of that constant effort & gratification! However, you've researched & rejected this...

A world-class full-range boxed speaker is, IMO, difficult to get right. Difficult means, it will take a huge investment in time & money & research (which means time & money, again). I wouldn't put it past your capabilities; just that the outlay could become large just by trying to correct common errors along the way. Quite a few parametres there, even without the passive filter. I'll give you two: which drive units and in what type of cabinet, shape, material:confused: Why??

Stats: as you imply a great stat (say a Soundlab) is difficult to build with the tools usually available in our sheds. It's also less well documented even though there are some kits available. Holland seems the place to go for these!

OTOH, it's difficult to get away from the "open" sound of a dipole (especially for classical)

The Orion -- contrary to yr belief, S Linkwitz has researched his drive units quite extensively for the Orion (that includes Thiel & partner's ceramic units). It's not a "cheap" speaker to build, but it's difficult to tweak (change amps?). I'd keep it on my list, especially with your budget!

X-over points & 4way. Not necessarily. I would try to keep the main (i.e. down to ~150-200Hz) as simple as possible. It really depends on the design & the units chosen. Wouldn't it be nice to be able to reproduce ~200-- ~5kHz with just one unit? A Tad tweet can then take you from ~5kHz to 120kHz (but blending drive units at 5-10kHz is a tough call:xeye: ). So that's only two-way down to 150-200Hz, even if there's narrowing dispersion in the midrange. Or, get a tweet that goes down lower (a Raven 3?). Then, the bass driver(s) can tackle the 150-40 range. For that last 1 octave, I would use another bass driver.

Another example: a front-loaded Lowther EX4 with tractrix horn measured inside a 12 db range (+/-5 in fact, with some peaks) from 18kHz--150 (@ 2m). Open baffle subwoof takes it from there to around 35Hz...

Too many considerations
:bigeyes:
Give a kit -- or a existing design -- a chance;) Cheers
 
I've the same goal, build a reference system. I believe it will be 3-Way with active crossover, I've the DSP behringer DCX2496, i'm going to buy a ROTEL RMB1066.
I believe a reference speaker is a high efficiency one for high dynamic, so goal is a 100 dB/w/m sensitivity between 150-20.000 Hz. It seems easily reachable without building a big box, two 10'' drivers and a 1'' compression in a bi radial Horn.

-B&C DE25 1'' dirver + his ME45 Horn is known for great linearity between 1K et 18K with a soft sound (mylar dome)
-B&C 10MD26 is also quite linear and will be great for 150-2000 Hz

For the bass, it's quite a big problem, there's no high efficiency solution that can enter in my living room and reach such low frequency than 16 or 20 Hz...

Two options :
1-great clean fast and linear bass between 50 and 150 Hz with a good 15'' pro driver (JBL, Beyma etc.) in a reasonable box
2-big low bass with a low efficiency speaker like the known peerless XLS and Shiva

I know the option number 2 beacause I've build small box with peerless XLS and deep bass are good enough, it's sometimes nice to go so low than 25Hz in frequency but bass are not so good between 50 and 100 Hz, that's a pity !
 
My take on a reference grade semi-active reference grade speaker

Konnichiwa,

The following represents my current thinking on "reference grade", which has developed somwhat, but surprisingly little, compared to my ideas nearly 5 Years ago.

Back in March 99 I posted my "A modest proposal" on the then Gizmo/Triode Guild board, I re-post in another psost for historical interest....

Here my take on "A modest proposal for a 21st century reference grade speaker"....

BTW, my term "A modest proposal" is in honour of Jonathan Swifts 1729 "A MODEST PROPOSAL FOR PREVENTING THE CHILDREN OF POOR PEOPLE IN IRELAND FROM BEING ABURDEN TO THEIR PARENTS OR COUNTRY, AND FOR MAKING THEM BENEFICIAL TO THE PUBLIC", the reading of which I strongly suggest to the gallant reader, as much said there still applies nearly rs later if not specifically to Ireland, but the world as large....

Anyway, back to speakers.

Starting from the Extreme low end, frequencies below the modal range of the room are best reproduced by a subwoofer system offering an omnidirectional dispersion, in other words a sealed box,. If we utilise a very large diameter pro audio woofer driver (18" or larger up to 24") in a small sealed enclosure, actively equalised and actively crossed over to the higher range system, we can move a lot of air at very low frequencies, giving 32' stopped organ pipes and the like good justice, with potentially structurally damaging Sound Levels down to < 10Hz.

At a point determined by the longest room dimension (set to a frequency slightly below this) we will start to employ a dipole woofer to cover part of the modal range of the room. It has been convincingly shown that dipole woofers operate much more consistently and evenly in acoustically small room than omnidirectional woofers.

Dipoles must be equalised against dipole losses and again need to move loads of air, so adding a second of our large Pro-Woofers (18"-24") in a dipole will be fine. This will operate into the around 100Hz range and take care of all the "high energy" parts of music, with the LF part of Tympani, Bass etc handeled here and by the superwoofer.

We can now use a stereo Tripath Module of the largest power level to drive each driver, use a swiutchable 24/28/32/36/40/44 Hz 4th order crossover between the two sections, suitably equalised and arranged.

The dipole Woofer can be placed above the main Fullrange Array and the sealed woofer below, making for a tall and narrow shape. With over a kilowatt per driver and 18" or larger long throw driver we should remain safely outside any limitations of the dynamic range.

As main drive array I would now recommend a large diameter (12" or 15") coaxial in order to assure suitable patter control, with the addition of a spherical waveguide to carry down this pattern control to as low a frequency as possible. The driver should be very low Qt and allow on an open baffle with a 24" Diameter Waveguide and a 1st Order passive Highpass on the Woofer a -6db point of 125Hz, to match in with our dipole woofer. This driver will again be dipolar, with an suitable lowpass filter behind it to suppress the rear radiation above around 250Hz (the delimting point between modal and diffuse range in most normal rooms).

A passive crossover, fully compensated for impedance is applied to this coaxial Driver (preferred choice Tannoy Alnico Coax Drivers such as Monitor Red, Gold or HDP or the modern Tannoy Alnico Drivers used in the Prestige series and available expensively but in excellent quality as spares) is to be used. The Coaxial Driver is set up to cover 125Hz - 12,500Hz with a 1,250Hz X-Over (all frequencies appx., the actual X-Over Frequencies I'd use are derived from a mixture of C37 and occult math).

A suitable supertweeter (most likely Horn - Fiostex, Visaton come to mind, certain ribbons also get a look in), takes over at 12,500Hz and extends the Range to 40KHz, so the whole speaker will manage <10Hz->40KHz as covered frequency range, with a sensitivity above the 95db/2.83V/1m range for the passively driven part and a power handling in excess of 100W RMS if the modern Alnico Tannoys are used, or in other words > 115db/1m per speaker in the 125Hz-12K5Hz range or for 3m listening distance in normal rooms around >112db RMS at the listening position.

If I where Douglas Self I'd likely call this project the "blameless" speaker, but being a man of modestly I prefer to call a speaker sporting a pair of 24" Woofers, a 15" lower Midrange, a 2" Voicecoil Compression Driver as upper midrange and a supertweeter oer side with housewreaker LF SPL potential "A Modest Proposal" (you can imagine what my "Immodest Prosal" looks like). Especially considering that it can be a pretty compact speaker by High End "Cost/Size No Object" standards, with a round 25" Wide and 75" high and maybe 18" deep, plus comparably unfussy to place.

Sayonara
 
A modest Proposal in 1999...

A modest Proposal (SEDHT-friendly Speakers)
From: Peter Pan (Thorsten@tnt-audio.com)
Date: 02 Mar 1999
Time: 11:57:18
Remote Name: netway-nhs.ukcore.bt.net


Comments
Hi all,

This one had me thinking for a good while. I hope to eventually write about it in the Main Pages of the Guild Web-Site, but for now I'd just like to open this discussion.

Now here is my "modest Proposal":

The proposed Speaker borrows heavily from the Legacy "Whisper", but attempts a different approach to resolving some of the issues.

http://www.legacy-audio.com/whisperpg.htm

The Unit I propose would use a wideband midrange array made up from the Audax Cabon-fibre Coned 5.25" Drivers.

These are high sensitivity (90db/W) and wide Bandwidth (as much as 60Hz - 10kHz). They are used in the "Von Schweikert" VR Series of Speakers and are indeed among the most natural and uncoloured sounding Midranges in current Production.

By using four of these Drivers in Series/Parallel Connection we achieve a Sensitivity of 96db/W with a relatively nice 8 Ohm Impedance. Cost is modest.

If we mount these Drivers onto a Baffle of about 17-20" Width we achieve a -3db point of about 200 to 160 Hz, below which the Response will fall off by 6db/Octave down to about 40Hz.

This Midrange Array will determine the overall Sensitivity of the Speaker, as well as the nominal Load impedance. By using a suitable mounting Arrangement (similar to Whisper) a Radiationpattern of about 50 degrees (horizontal) and 70 degrees (Vertical) is achieved across the whole Midrange for frontal radiation.

The Rear-Radiation will follow similar Pattern alowing for simple measures to disperse this wave diffusely or use it purposly to enhance the Airyness of the percieved sound.

As the Drivers are mounted only on a flat Baffle, there is no Cabinet or Horn to colour the sound, to store energy and to radiate the backwave back through the cone.

The Problems will start by the Time we try to match a Bass-System and a Treble Unit to our 200 to 160Hz - 8kHz Midrange array.

The solution used by Whisper that essentially uses two complete Pannels mounted behind each other and uses overall 4 X 15" Drivers is essential in achieving a good and matching Bass.

To do some Calculations, if we assume that we wish to extend the lower -3db Point from 200Hz (160Hz) to 50Hz (40Hz), we need to apply a 6db per octave Lowpass to the Bass Array.

We will overall require an additional 12db Sensitivity to match the Midranges 96db/W. If we use four 15" Drivers of a Type capable of 101db/W we will have essentially achieved that Aim, as the resulting Sensitivity will be 107db/W, 11db above the 96db/W.

A range of high Quality Pro-Audio 15" Drivers will be suitable. I would look for units having really LARGE Magnets and cast Frames. But I think even the relatively cheap Eminence Units will work reasonably well in this Application.

In addition to the above, if a 1st Order Lowpass is applied to the Bass Units, the Cross-over between the Bass Array and the Midrange Array will be seamless and determined by the baffle.

It would be recommended to use a suitable Series Capacitor for the Midrange Units to reduce the Cone Excursion at high levels of low frequencies. However, with the Audax Units this can be omitted, as long as the Input Power is being kept below 10-20W (which will produce 106-109db anyway).

In any Case, if a 20" Wide Baffle is used, the Speaker will be essentially Flat down to about 40Hz with a very gentle rolloff below this frequency, giving a -9db Point of about 20Hz (depends somewhat on the Bass-Units).

The beauty of a Dipole (Sub) Woofer is that it interacts a lot less with the room and hence will reduce the degree by which room-modes are excited quite a bit.

So, now we have a modestly large, fairly easy to place Speaker with 40Hz - 10kHz as -3db Points (and I mean -3db not +/-3d, not the ofthen quoted -12 or -20db "Frequency Range" or whatever.

Real hard Hz's and db's. The Tweeter will be a bit of a problem. My first chioce would likely be the Focal "Telar" Driver. With that extremely expensive and hard to obtain, a professional "Slot" Radiator, or indeed a "Bullet" could be used.

In either case the match of Directionality will be sub-Ideal.

My preferred solution would be a Slot-Radiator, turened in such a Way that the "Slot" is almost Horizontal (as opposed to the usual Vertical mounting), but turned of square by 15 - 25 Degrees so that inner parts are raised.

While this will give a VERY wiered Radiation Pattern, it will be one that will work (IMHO) well in most conventional living rooms.

Due to the often high Powerhandling and reasonable resonance Frequency with these Slot Radiators (JBL 2405 comes to mind) the X-Over from the Midrange Array to the Tweeter is unproblematic.

Depending upon the type and Make of Slot Radiator a 4-10db attaenuation will have to be applied. After that and accounting for the Input Power-Limit of 10-20W RMS, the X-Over can be as little as one single good Quality Capacitor, with the X-Over frequency set to about 8 - 12kHz.

In order to make a Baffle that stores little Energy bit is rigid, I suggest a Sanwich made from a 1-2" Styrophor Foam Core, covered on all sides by about 0.5" pre-veneered MDF, laminated together with a glue that remains slightly tacky.

A very heavy base that is rigidly attached to the Pannels will be neccesary. Something suitable build from the same veneered MDF and having a large Lead-Shot & Sandfilled Cavity will anwer the purpose perfectly well.

Alltogether the resulting Speaker should be possible to assemble for less than $ 2,000, is realtively easy to assemble and place.

And it offers a simple, easy to drive 8 Ohm load with a genuine sensitivity of 96db for 1 electrical Watt.

Now all that remains is to find a company selling Drive-Units (or Audax) to sponsor the development of this Monster.

Okay - the thoughts so far. Over to you Ladies & Gentlefolks. Tell me what you think.

Later Peter Pan
 
HiFiNutNut said:
It surely can. However, if we use active XO, I believe that 4 way will be far superior

4 way active is superior, I will agree.

However, it adds needlessly to your cost, size and complexity.

Provided that you can do so with your final design _grin_ I suggest you consider three way active - using a tweeter with only a cap on top -

This will allow you more options with a digital crossover and would allow you to use the Behringer or Driverack PA -

It also would allow you to allocate more money to other parts of the system


peranders said:


Short posts are golden.

As to the rest of the thread - My ADD kicks in _grin_
 
"Phase Alignment

I believe that the active, LR phase-linear XO will take care of the issues. Because I use active XO, I have avoided all the phase problems associated most passive XOs."

That is simply not true. Active crossover has the same phase characteristics as a passive one given the same transfer function. A Linkwitz/Reiley crossover is a Linkwitz/Reiley crossover regardless of wheter it is performed on line level or speaker level signals.

If you're going for a 4 way "cost is no object" type speaker, at least pay attention to correct wave form reproduction, which means a transient perfect design. You can achieve it by going with all first order acoustical slopes or by employing digital crossovers with proper algorithms that allow for steeper slopes while preserving the transients. If that is what you meant by "active phase linear" then remove the "LR" from the description. On another related note. Do not believe that time domain is less important or secondary to frequency domain. A trained ear can easilly hear the difference between two speakers where the transient reproduction is the only variable. I designed and participated in such listening tests. The extra coherence is well worth the effort.

"I am not so sure about time alignment. Some people think it is not important at all and it is only for marketing purposes. I think I will initially build separate boxes so that I can physically align them and see if it will make any difference."

Time alignment is a crucial part of the crossover design. When you move drivers back and forth the frequency response of the system changes due to phase differences of multiple signal sources.

"As for diffraction, I am not sure about the actual impacts. Some people say it is very important. Some people doubt it. I guess that I may try to round up the box edge / corner, if possible."

If you design your crossovers based on measurements taken on a baffle with sharp corners then that's what you got- a speaker optimized for a particular enclosure. Just don't make a crossover for sharp corners and then round them without remeasuring and adjusting the crossovers accordingly. The sharp baffles induce more of a ripple effect on frequency response. You can correct them with active filters, but it's better to have no ripples (or less ripples) to begin with.

All in all, based on your post, I would advise you to build someone elses proven design (such as the Orion plus sub). With a 4 way there is way too many variables for you to get it just right. If you're set on creating something new and unique, you might want to consider cooperating with someone experienced and equipped to tackle such a big project.
 
Thunau said:
"Phase Alignment

I believe that the active, LR phase-linear XO will take care of the issues. Because I use active XO, I have avoided all the phase problems associated most passive XOs."

That is simply not true. Active crossover has the same phase characteristics as a passive one given the same transfer function. A Linkwitz/Reiley crossover is a Linkwitz/Reiley crossover regardless of wheter it is performed on line level or speaker level signals.

I agree that the DSP will not solve automatically all crossover complexity...What is method to get good phase and transient response ?
I was planning to feed the loudspeakers with the crossover frequency and to move phase until SPL is max, doing this for each frequency corner will give good result, no ? And what about transient response, it depend more of the box than crossover, no ? Can you explain please...
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.