Opinions on this design for desktop speaker?

[10PSPMemSticksForMiku2007]

I've been trying many different desktop speakers and am not satisfied with any of them, so I came up with an unusual design. Attached is a crude diagram, not at all to scale, but it illustrates the idea.

I want to achieve the following - flat response from at least 40hz to 17khz with little distortion, point source or close to it, narrow radiation, excellent dynamics, good power efficiency. Despite having good bass, I don't want it to vibrate my desk. With the PC, I have 8 channels output and unlimited DSP. The desk is quite big.

I plan to make a WAW with slot-loaded woofers and a horn centered between them. It will fill the vertical space between my desk and ceiling to get more internal volume than normal desktop speakers. Each speaker is split into two sealed boxes with a common rear panel and a slot between them. Roughly 8" wide x 12" deep x 42" high or ~2.3 cubic feet for each half. In the slot, 4 12" woofers in push-pull. Each pair of will be as close as I can fit them with a compression driver + horn in the middle. The width of the slot would be about 5 or 6 inches, as small as I can make it. So each speaker is about 24" wide in total.

For me, the huge size is no problem, but it would be possible to create a much more reasonably sized version if you don't want so much output at such low frequencies.

For the CD I'm thinking of the 1.4" Eminence N314T-8. Horn is FaitalPro LTH142 60x50 tractrix which claims linear loading to 500hz. Woofers, GRS 12PT-8. Parts cost is $600 for each speaker, not counting the cabinet.

With DSP I should be able to get a flat response to 30hz at 107db at 1 meter. The crossover point and slope is completely flexible but I'm hoping around 500hz. Sitting only 1 meter away I hope I'm not torturing the CD too much by doing this. The other concern is high frequency cancellation from slot loading the woofers, especially if I need a higher crossover. Based on this thread about a slot-loaded 12" w-baffle subwoofer, it looks like it would barely work.

I'm not planning on a unity horn because I absolutely want the woofers in a force-cancelling arrangement. I'm only hoping the difference in radiation pattern does not matter so much if I can get a very low crossover.

So far, it is looking OK to me, but this will be the first time I use a compression driver or use slot-loaded woofers to such a high frequency. It would be really great to get feedback from someone experienced in these topics before I spend $1000+ on this. In that case, thank you in advance for taking the time.

 

[Kalasbarnet]

i voted yes because i think it is a cool idea but i have no real knowledge lol! i have been pondering something similar but with the horn+CD substituted for a MEH and the woofers doing sub duty

 

[kemmler3D]

I'm only hoping the difference in radiation pattern does not matter so much if I can get a very low crossover.

Yes, this was my first thought when I saw you intend to put 12" drivers on a desk. My non quantified guess is you will have problems with summation even with a low crossover...

On the other hand we're pretty much talking about nearfield subs here, which I understand is a good way to avoid room modes.

So maybe it's actually a great idea? I'm just not sure whether you will be sitting far enough away for it to sound right.

 

[10PSPMemSticksForMiku2007]

Thanks for the input. I guess I really don't know what this will sound like unless I try it, and I should try it before I buy it.
I have a couple sealed 8" woofers and small satellite speakers that I can use for prototyping with the slot depth and center to center distance - which ideally I could get to 6" with 12" woofers, although I might have to give up on push-pull if I do that. Even with the woofers positioned like that, the horn is about 10" tall, so the woofers would be almost halfway obscured behind the horn. I have a feeling it will work okay.
500hz is an octave higher than you would cross mids to door/kick woofer in a car, yet the distance is much greater there, and it can sound good with time alignment.

 

[Duke]

@10PSPMemSticksForMiku2007, I admire your creativity but imo the slot-loaded woofer configuration presents some issues.

Even if you were able to get away with a 500 Hz crossover frequency (unlikely with that horn and compression driver combination if high SPL is a requirement), the far edge of the woofer cone is nearly 1/2 wavelength further away than the near edge in the 500 Hz ballpark. This will cause some cancellation, but it can probably be compensated for with EQ.

This next issue cannot be fixed with EQ: The arrival time of the cones' output will be smeared by a significant fraction of a wavelength, moreso at the upper end of the woofers' range than at the lower end. EQ cannot fix that time smear. How audibly significant would it be? I don't know, but I've heard imo unacceptable degradation of clarity from less time smearing in that frequency region. And there will also be a reflection off the surface at the innermost end of the slot.

Then there will be other resonances within the slot itself. Their effects on the frequency response can probably be largely compensated for, but I'm not sure about their effects on the time-domain behavior.

I hate offering "criticism" without also offering a potentially useful suggestion.

You might consider an MHM configuration with the horn's wide dimension in the horizontal plane and then angle the bottom woofer up a bit and the top woofer down a bit such that their axes are converging, and then angle the whole array downwards so that it's "aimed" at you, or at least "aimed" at your ear height. When you are in your designated listening location, you want the upper woofer and lower woofer to both be the same distance from you or at least very close to the same; imo you don't want two different arrival times for the woofers' portion of the midrange region if you can help it.

So, perhaps something conceptually similar to this B&C design:

 

[ThatSoundsGood]

I really think 500Hz is too low for that CD (or just about any 1.4" CD). I wouldn't go lower than the recommended 800Hz. Maybe you can get some loading down there with the waveguide and maybe it will be fine if you don't play it loud but that's a lot lower than what is recommended. With DSP you could try a steeper slope (maybe 24db/oct or 36db/oct) around 500Hz and see what the distortion measurements are but you're definitely going to be pushing it.

 

[Arindal]

Sitting only 1 meter away I hope I'm not torturing the CD too much by doing this.

If you are that close to the speaker, I would really make sure you don´t run into issues with frequency response changing slightly at different angles when moving your head. Difficult to judge simply based on measurements, so I would recommend to check the horns plus conventional midwoofers in a listening comparison before actually building the 4x12" cabinets. When moving your head in the nearfield, a horn of that size could produce all sorts of unwanted effects, as well as leading to an overly direct projection plane which can be annoying with certain recordings.

Such horns are not really made for true nearfield listening.

As Duke has mentioned, I would be concerned that the distance between the two woofer arrays might be a bit overstretched for a x-over point at just 500Hz. I would not dare to predict anything, but you might want to make sure you don´t get issues with imperfect vertical localization, lobing and cancellation. Even more so at higher x-over points like 800Hz.

To prevent all these issues, a coaxial arrangement of a horn plus a single 12" midwoofer might be a good idea, accompanied by a pair of additional 12" subwoofers in opposite direction, in one and the same enclosure which you can run as a 2.5way concept. Much more promising for nearfield in my understanding.

 

[10PSPMemSticksForMiku2007]

Thanks again for your feedback everyone.

This next issue cannot be fixed with EQ: The arrival time of the cones' output will be smeared by a significant fraction of a wavelength, moreso at the upper end of the woofers' range than at the lower end. EQ cannot fix that time smear. How audibly significant would it be? I don't know, but I've heard imo unacceptable degradation of clarity from less time smearing in that frequency region. And there will also be a reflection off the surface at the innermost end of the slot.

Thank you for your detailed explanation and alternative suggestions. As I did a bit of prototyping and listening tests, I think you're right about this. So I need to either change the design completely or make it a three-way - and even then the mids would be too far for the higher crossover, which defeats the purpose. I would also prefer my new speakers to be two-way so I have channels left for rears.
I'm also still not sure a unity horn would make sense. At the least it would have to be narrower than the other ones I've seen which would make it hard to build and fit where I need it to.
So, maybe something like you suggested, but maybe also something like this? I know that with unity horns it is common to have woofers partially, even mostly, obscured behind slots. Because of the horn's dimensions, there's a limit to what center to center distance could be achieved. The B&C design you posted looks very nice, but I don't know how coherent it is at 1 meter.
Next I will prototype this layout with the same center to center distance and crossover frequencies although I have 8" woofers to work with now. This will give up some output but it's a simpler direct radiating design. 2 10s, or 2 12s, or I could even do 4 10s surrounding the horn if I can still find a way to mount the horn around them.

On the other hand we're pretty much talking about nearfield subs here, which I understand is a good way to avoid room modes.

I'm currently using stereo subs about 1m away, directly next to each speaker. There are still a lot of room modes involved, seemingly as much as other placements I've tried. I think you get volume by sitting closely but the frequency response still needs correction.

unwanted effects

What would be the unwanted effects other than very narrow directivity (which I do want)?
Also, to me it seems having a crossover around 500hz to 1000hz with 5" or 6" c2c distance would be proportionally no worse, maybe better, than your typical tweeter being 3+ inches from the mids. I don't like to have that crossover around 2-4 khz when sitting so close to the speakers.

As for force cancellation, I think it would help, but metal desks vibrate with only acoustic energy despite being solidly built, and I think I have to put resonix on the panels no matter what, but I was hoping that I don't have to make it look ugly.

I really think 500Hz is too low for that CD (or just about any 1.4" CD). I wouldn't go lower than the recommended 800Hz. Maybe you can get some loading down there with the waveguide and maybe it will be fine if you don't play it loud but that's a lot lower than what is recommended. With DSP you could try a steeper slope (maybe 24db/oct or 36db/oct) around 500Hz and see what the distortion measurements are but you're definitely going to be pushing it.

My plan was to buy the stuff, hope and pray I can get such a number, but the crossover point can always be adjusted. I don't really see anything else better let alone cost effective in this area that would go much lower. I could go up to the Celestion Axi2050 for $1000 a piece but the horn would be too small to really benefit from it, and that can't really be increased any more while still fitting on a desk. A coaxial + 12" with comparable horn still ends up with the horn covevering most of the woofer on axis, seems even worse than my pic. Aren't there dips and directivity errors at much higher frequncies with that design?

 

[Duke]

I could go up to the Celestion Axi2050

The Axi2050 has a 2" exit. The dual-diaphragm B&C DCX464 and DXC354 both have 1.4" exits, so they could be used with that Faital horn. But imo you wouldn't gain much low-end extension versus the HF1440 because the horn itself is the limiting factor, its mouth is too small for a crossover frequency much lower than about 800 Hz. Ime.

A coaxial + 12" with comparable horn still ends up with the horn covering most of the woofer on axis

Imo the preferred configuration for a coaxial is to use the woofer cone itself as the horn, and then cross over high enough that the woofer cone is just barely moving. Again Faital comes to mind; you'd have to do some EQing:

https://faitalpro.com/en/products/Coaxial_Loudspeakers/product_details/datasheet.php?id=351050020

 

[10PSPMemSticksForMiku2007]

Imo the preferred configuration for a coaxial is to use the woofer cone itself as the horn, and then cross over high enough that the woofer cone is just barely moving. Again Faital comes to mind; you'd have to do some EQing:

https://faitalpro.com/en/products/Coaxial_Loudspeakers/product_details/datasheet.php?id=351050020

A three-way system then. What crossover would you suggest for the coaxial mid/woofer?

 

[Ktacos]

In traditional DIY fashion, many suggestions on what to make and what not make have been made, but no one has asked why all the speakers you've tried out don't work. Desktop environment doesn't need 107db at any frequency, and compression drivers and horns tend to not really sound nearly as good as a good dome/wg combo. The high efficiency isn't needed here.

There are a lot of excellent speakers that will give you state of the art performance at a desk, they're not even expensive, dimishing returns kick in super fast for nearfield desktop setups. If point source is a goal of the project I wouldn't even look at DIY, none of the coax drivers out there are as good as what you can get from KEF.

 

[Duke]

A three-way system then. What crossover would you suggest for the coaxial mid/woofer?

I haven't studied this configuration in-depth enough to have a well thought-out suggestion, but off the top of my head, for a high-SPL application I think the crossover between the coaxial "midrange" cone and your woofer(s) should be as high as you can reasonably get away with. You mentioned 500 Hz earlier in the thread, and I think that would probably work well. Imo you want at least a 2nd order highpass slope, as with a 1st order slope the cone excursion continues to increase as we go down in frequency even though the actual sound pressure level is decreasing.

That particular Faital coaxial has about two octaves of overlap between the cone driver and the compression driver, giving you lots of options for your crossover frequency between coaxial cone and compression driver.

Note that with a round horn, and therefore with this type of coaxial, there will be a dip in the upper range of the on-axis frequency response. What happens is this: There will be one path length from the compression driver directly to your ears, and there will be another path length from the compression driver to the outer edge of the cone, where the sound is reflected, part of that reflection continuing on to your ears. When you are directly on-axis that reflection arrives coherently from all around the perimeter of the cone, and there will be a frequency where the difference between those two path lengths is equal to one wavelength, causing cancellation. There will be a dip centered on that frequency. The solution is to listen from about 15 or 20 degrees off-axis, so that the reflection arrives somewhat incoherently (spread out in time), causing little or no such dippage. In other words, if you're using coaxials or round horns, particularly in a nearfield setup, don't aim them at your ears.

Just about everything's a tradeoff in speaker design, and for a custom design like yours, without having a large R&D budget you have to make your best guesses about a lot of things. I can see a coaxial making a lot of sense for what sounds to me like a giant nearfield speaker.

 

[10PSPMemSticksForMiku2007]

In traditional DIY fashion, many suggestions on what to make and what not make have been made, but no one has asked why all the speakers you've tried out don't work. Desktop environment doesn't need 107db at any frequency, and compression drivers and horns tend to not really sound nearly as good as a good dome/wg combo. The high efficiency isn't needed here.

There are a lot of excellent speakers that will give you state of the art performance at a desk, they're not even expensive, dimishing returns kick in super fast for nearfield desktop setups. If point source is a goal of the project I wouldn't even look at DIY, none of the coax drivers out there are as good as what you can get from KEF.

As I said, I want narrow radiation, imaging, and good dynamics with low distortion 40hz and above. I don't plan on actually blowing my eardrums out here, but the ability to be maintain clarity during a loud moment is desired.

I currently have typical 3-ways with 8" woofers and fully DSP crossover. I was never satisfied with the woofer/mid blending with the tweeter in any bookshelf speaker at 1 meter. I can easily hear the high and low notes coming from different heights, and the texture in the crossover region is never quite right despite that I can make sure it is perfectly time aligned and try any kind of crossover I want. Even with 3" mid barely 1" from the tweeter - their outer mounting rings are touching - it is not transparent at this very short distance. I often remember how my old basic computer speakers sounded better in this regard, so I want to try a wideband again. Compression driver + horn seems better to me than the "full range" options right now. Coaxials are also options I am skeptically optimistic about.

A lot of the commercial bookshelf speakers and studio monitors are good for their size, amazing even, but that size still comes with limitations in bass performance. Even highly rated, expensive ones do not have such good distortion numbers from ~50hz even up to 200hz in many cases. If they actually do have good output and low distortion and 40hz it comes with group delay. If that has been fixed with DSP it comes with high $ cost and latency. They are just too small drivers in small boxes. I don't want a monitor with just one 5" woofer. I can easily and cheaply just make them bigger. Most big speakers are not designed for close listening, but they could be.

That said, it is an option to buy KEF speakers and sandwich it in between some woofer boxes. Easy enough to cross that at ~250hz. It would be a hybrid crossover but still require only 2 DSP + amp channels. That did actually occur to me before posting this thread. I went to Best Buy and listened to some KEF bookshelf speakers there, but I wasn't able to audition to them in the way I wanted to. Also, they are not that much cheaper. $800 a pair for basic model, then I have to add real woofers as well. BTW, the satellite speaker I used for testing earlier is an older 3" KEF unit. Pretty good for a sattlite speaker I guess, but I'm not amazed with it. There's a huge dip around 2 khz that cannot be eq'd. Maybe a better crossover is needed.

While it's not necessarily a bad idea, I see all this stuff on Parts Express with similar cost and incomparably more powerful. I really like the idea that it can be arranged into a super "nearfield" speaker. Also, this is a hobby in addition to pragmatic problem solving, so buying a KEF speaker and being satisfied is not really a hobby activity to me.

Just about everything's a tradeoff in speaker design, and for a custom design like yours, without having a large R&D budget you have to make your best guesses about a lot of things. I can see a coaxial making a lot of sense for what sounds to me like a giant nearfield speaker.

I appreciate the recommendation for that driver, and I also have been looking at other Faitalpros. However, I questioned why I would spend $400 to $1000 on a woofer that works as a worse horn only to cross it at 500hz, when I could use an actual horn and cross at 800hz surely, 500 if I'm lucky.
Looking into it, the "rule of thumb" I've found is to keep the woofer under .5mm excursion. So that's great news, you can do a much lower crossover. Low enough that you can do front-firing coax and side-firing woofers on a large enough box!
It follows that bigger woofers have lower excursion so you want the biggest one for that. 15 or 12 surely. I don't know if this comes from pro circles or hifi, but if we go with that, the 15HX500 could be crossed as low as 100hz @ 10 watts. Might be promising but when you subtract the chamber size the coaxial requires, there is not much left for bass in terms of either volume or physical placement to get the coax centered between them. Yet if you had more space, you'd probably get a smaller coaxial and cross with front-firing woofers.
I think these could potentially work and sound great in some applications, yet don't really function full-range on their own, yet are quite more expensive than what I was planning. I should get similar cone area and much less concern of intermodulation distortion if I do the MTM design above. I'm not really sure what benefits I'd get from going three-way in this case.

 

[Arindal]

What would be the unwanted effects other than very narrow directivity (which I do want)?

The main issue with several separated sources in a nearfield environment is cancellation effects between drivers the moment you slightly move your head, leading to weirdly changing tonality. As well as uneven tonality over the listening window which effects perfection through HRTF. It simply is not uniform but you have peaks and dips, particularly if you use a tweeter horn in close proximity which it is not meant for.

A side effect of this is unstable localization, like phantom sources ´wandering´, jumping or getting blurred, as well as a certain degree of subjective ´phaseyness´. The wider sound sources are apart, even vertically, the more blurred and non-homogeneous your imaging will become. It kind of contradicts your goal go stable localization IMHO.

The question would be why you want even narrower directivity? In a nearfield setup, directivity is usually regarded to be narrow enough the moment you cannot perceive the effect of indirect sound/reflections, which at 1m is pretty easy to achieve. Neither proximity nor localization stability cannot be improved further from this point, so I would suggest to focus on other goals to help localization and imaging, such as moving all sound sources together (ideally coaxial), using smaller drivers (a 12" midrange as seen from 1m is much broader than a phantom source should appear) and do everything for uniformity of frequency response over a vast window.

I was never satisfied with the woofer/mid blending with the tweeter in any bookshelf speaker at 1 meter. I can easily hear the high and low notes coming from different heights, and the texture in the crossover region is never quite right despite that I can make sure it is perfectly time aligned and try any kind of crossover I want.

Another way of expressing what I tried to describe from studio monitoring point of view. My concern would be this gets worse with the concept in question, not better.

Even with 3" mid barely 1" from the tweeter - their outer mounting rings are touching - it is not transparent at this very short distance.

If you really get the ideal listening plane vertically, you might be able to avoid such problems, but I agree that even such a concept is not ideal under extreme nearfield conditions.

If you are sensitive to such phenomena, the only solution I see is using any kind of coaxial concept in the mid/high region, preferably a more compact coaxial driver offering the maximum of uniformity in frequency response over different angles (which is not easy to find on the market). Or some medium-sized coaxial for which additional bass units are available so you can make it a 2.5-way concept with 2 amp channels, the woofers get a passive 1st order low-pass filter and that is it.

A lot of the commercial bookshelf speakers and studio monitors are good for their size, amazing even, but that size still comes with limitations in bass performance. Even highly rated, expensive ones do not have such good distortion numbers from ~50hz even up to 200hz in many cases. If they actually do have good output and low distortion and 40hz it comes with group delay.

I agree on the limitations when it comes to subjective listening tests, but I would not overly worry regarding THD and group delay. If a capable speaker is properly designed and operated within reasonable SPL limits under nearfield conditions, you would not have problems with audible THD or group delay.

If you want a reference in this regard, try to listen to KSD C120 or C100 Reference monitors. They are exceptional both in terms of bass quality and group delay as they are linear phase.

I questioned why I would spend $400 to $1000 on a woofer that works as a worse horn only to cross it at 500hz, when I could use an actual horn and cross at 800hz surely

Because any larger horn known to me which would meet these demands, is not near field-capable in terms of uniformity @1m, and you would face huge difficulties combining it with a midwoofer without localization and cancellation issues. There is a reason why manufacturers of large, horn-loaded coaxial drivers put a lot of effort into R&D and charge a significant amount of money for those. It is really difficult to design them properly. You might want to check Beyma, B&C and LaVoce, except from Faitalpro, as these are all really specialists in designing such drivers and lauded by many speaker developers.

Looking into it, the "rule of thumb" I've found is to keep the woofer under .5mm excursion.

For horn-loaded coaxials or such with larger diameter of the tweeter unit inside the midrange cone or tweeter being placed in front of cone, I would not consider this to be critical, as the movement of the cone hardly affects the treble dispersion. The rule of thumb in my understanding is applicable to conventional coaxial units as known from Genelec, KEF, Tangband or Seas.

All the best for your project!

 

[Ktacos]

As I said, I want narrow radiation, imaging, and good dynamics with low distortion 40hz and above. I don't plan on actually blowing my eardrums out here, but the ability to be maintain clarity during a loud moment is desired.

So many speakers will do this just fine at 1 meter.

I currently have typical 3-ways with 8" woofers and fully DSP crossover. I was never satisfied with the woofer/mid blending with the tweeter in any bookshelf speaker at 1 meter. I can easily hear the high and low notes coming from different heights, and the texture in the crossover region is never quite right despite that I can make sure it is perfectly time aligned and try any kind of crossover I want. Even with 3" mid barely 1" from the tweeter - their outer mounting rings are touching - it is not transparent at this very short distance. I often remember how my old basic computer speakers sounded better in this regard, so I want to try a wideband again. Compression driver + horn seems better to me than the "full range" options right now. Coaxials are also options I am skeptically optimistic about.

No offense but there's no such thing as typical 3 way, they're all different. If your filtering is truly well done, you shouldn't have much issue with localizing the drivers. Perhaps you have diffraction issues, idk I have no idea what this speaker looks like. The fact that you're stating "I can try any crossover I want" makes me wonder just how well the drivers are integrated. Flanges being close isn't enough information, some drivers have massive flanges in relation to their diaphragm. The old computer speakers comment feels like nostalgia.

While it's not necessarily a bad idea, I see all this stuff on Parts Express with similar cost and incomparably more powerful. I really like the idea that it can be arranged into a super "nearfield" speaker. Also, this is a hobby in addition to pragmatic problem solving, so buying a KEF speaker and being satisfied is not really a hobby activity to me.

Just buy some sica 5.5 and a woofer to go with it and make a 3 way.

A lot of the commercial bookshelf speakers and studio monitors are good for their size, amazing even, but that size still comes with limitations in bass performance.

Just use subs. Trying to get everything to come from one place is a recipe for disappointing bass. If you want excellent bass in room, multisub is the answer.

Coaxials are also options I am skeptically optimistic about.

Wasn't one of your goals in the OP point source? And instead of a coax, your first consideration was a CD and horn? But you're skeptical of coax?!

 

[10PSPMemSticksForMiku2007]

I would like to thank everyone again for their input. I have been looking more into the pro coaxials with compression drivers. I was worried about these because of my experiences with the consumer coaxial speakers, but it seems these really are different and can even have distortion below 2% with 120db according to a review I read on AudioXpress for one of the B&C models. Unfortunately it seems thorough reviews of the newest models are relatively scarce, but I am looking at various 12 and 15" models. I think I will go with one of these. Most likely to build cabinets in the spring or summer.

 

[Ktacos]

PA coaxials typically mean sacrificing quite a bit of sound quality, they're just not nearly as refined as kef/genelec/sica. Their first priority is just getting loud. I mean just look at the manufacturer responses on them, they're a mess. DSP can't fix the problems they suffer from either.

 

[Flaesh]

sica

Aren't Sica PA?..

Mark @gnarly , should we change our coaxes to some KEFs?..))

 

[Ktacos]

Aren't Sica PA?..

The models that are actually good are mostly for home use.