Help with centre speaker design

[aamacgregor]

Problem
So I have a Definitive Audio Mythos Seven centre speaker, it's ok, but sonically it doesn't match in with my main speakers - to be honest, by comparison to my main speakers, the Mythos sounds compressed, closed in and a bit muffled.

Notional Solution
So, I had the great plan to build a centre speaker using the same drivers as my main 2-way speakers. The drivers in my main speakers are Scan Speak 18W/8545-00 and D2905/950000's.

I wanted to build an MTM centre speaker using the above drivers, but in order to reduce horizontal phase issues, I'm thinking of making it a 3-way W/M-T/W design, using the later 18W/8545-01's as the woofers, either side of a vertically aligned D2905/9500000 and ScanSpeak 10F/8424G00. I'm planning on crossing the W/M at 800Hz and the M/T at 3k in order to maintain a good off-axis response.

Concerns
1) I've never built a speaker before and am wondering if a 3 way is a bit ambitious. That said, I'm sure I'll learn plenty, even if there is plenty of tweaking that takes place.
2) A crossover frequency of 3k is possibly not ideal to cross at. I've heard some say avoid crossing between 2k to 3.5k, since it's not only in the presence region, but it is the region the ears are most sensitive to inconsistencies. From what I hear, if I cross in this region, I have to make sure the drivers are very well matched and that the cross over is bang on. Comments and suggestions on this point are very welcome.
3) The tweeters and mids are only sold in pairs (which means I'd have one left over of each - seems like a bit of a waste - anyone know if I could purchase these drivers as singles anywhere).

Questions
Does my suggested solution sound like a good plan? The mid is lower efficiency than the woofer and tweeter - any suggestions of reasonably priced drivers that would work better would be welcome. The thing is I'm not wanting the mid diameter to be more than 12cm (so the cabinet doesn't get too high).

I've been studying baffle edge diffraction and baffle step compensation... from what I've learned, I figure that 3cm or so radius edges of my main speakers probably isn't doing a whole lot. Do you guys worry about baffle edge diffraction, looking at the majority of commercial designs, it seems most manufacturers don't pay a lot of attention to it. Anyway, at the moment I'm planning on taking it into account as best as I can whilst trying to maintain a design that is aesthetically pleasing enough for my wife.

Test Equipment
I've already bought a Dayton Audio EMM-6 and Berhinger UMC202HD in preparation of making measurements.

 

[andy19191]

1) I've never built a speaker before and am wondering if a 3 way is a bit ambitious. That said, I'm sure I'll learn plenty, even if there is plenty of tweaking that takes place.

It depends on your capabilities and expectations. If the task is challenging the first effort is inevitably going to be less than ideal but this is not a problem if it is reasonably in line with your expectations and plans.

2) A crossover frequency of 3k is possibly not ideal to cross at. I've heard some say avoid crossing between 2k to 3.5k, since it's not only in the presence region, but it is the region the ears are most sensitive to inconsistencies.

A conventional 3 way is going to crossover in this region and well designed 3 ways sound good to my ears. If you push the crossover higher or lower you will almost certainly do more harm than good.

3) The tweeters and mids are only sold in pairs (which means I'd have one left over of each - seems like a bit of a waste - anyone know if I could purchase these drivers as singles anywhere).

By using a midrange you have lost sounding the same as the main speakers. The tweeter has a 4" face plate which is not normal in a 3 way centre speaker. A tweeter with a lot smaller front is likely to be a better choice. This will enable a more efficient midrange to be an option.

Does my suggested solution sound like a good plan? The mid is lower efficiency than the woofer and tweeter - any suggestions of reasonably priced drivers that would work better would be welcome. The thing is I'm not wanting the mid diameter to be more than 12cm (so the cabinet doesn't get too high).

There is the Audax HM100Z0 if you want a more efficient midrange. Another option is using the midrange as a filler driver with output from the woofer and tweeter making up for the lack of sensitivity in the midrange. Another is using a tweeter with a small face giving room for a larger midrange.

Do you guys worry about baffle edge diffraction, looking at the majority of commercial designs, it seems most manufacturers don't pay a lot of attention to it.

Baffle diffraction puts dips and peaks in the frequency response, the drivers have dips and peaks in their responses, the crossover puts dips and peaks in the response,... The better designs will juggle with all these to even out the frequency response but it may not always be apparent from looking at the finished speaker. If you look at the better commercial speaker designs these days there have clearly been efforts to reduce the diffraction from the woofers which is less of an option for DIYers.

 

[eriksquires]

Gravesen has a number of good center channel designs you might like here.

Your plan is good, but it's also quite ambitious. To go this far from scratch you need to learn how to analyze a speaker system and doing that well needs tools and education. If you want to get something quickly that works well, find a design from Gravesen or others.

You'll also need to analyze the driver's T/s parameters and the in-cabinet impedance curves. For that you'll need Dayton DATS V2 or similar.

Audiohobby.eu only sells in pairs, but Madisound sells singles. I think.

Lastly, because they are off-spec, getting good bass with the Scanspeak's requires a lot more volume than you would think. If you have it great, but otherwise your centers could be huge. SS is notorious in this line for being off spec. Look at Madisound's recommended cabinet sizes as a better guide than the published specs.

 

[jReave]

Two important consideration right off the bat:

Is your AVR happy with 4ohm nominal speakers? Running 2 x 8ohm woofers in parallel will give you 4ohm.

And what kind of placement are you looking at for the CC? If it sits on a small stand out in open space, it will see the normal 6dB of baffle step loss and will probably work as planned. If on the other hand, you are fitting it on a shelf in something like an entertainment center, then you may not really see much baffle step loss at all in which case 2 of the 8ohm woofers in parallel may be too high in SPL to match your mid. Even if you go with the 4ohm 10F4424.

Before purchasing anything, you should model and simulate your design. This will give you a better idea about feasibility.

Try Falcon Acoustics for your parts source.

 

[aamacgregor]

It depends on your capabilities and expectations. If the task is challenging the first effort is inevitably going to be less than ideal but this is not a problem if it is reasonably in line with your expectations and plans.

I'm happy with the thought of it not being overly straight forward and take a fair amount of effort. I figure I can reduce the number of cabinet iterations by simulating as much as I can, although I understand that the simulated values don't exactly fit the measurements, so it's possible I could end up building more than one cabinet in the end.

A conventional 3 way is going to crossover in this region and well designed 3 ways sound good to my ears. If you push the crossover higher or lower you will almost certainly do more harm than good.

That's good to know, thanks.

By using a midrange you have lost sounding the same as the main speakers. The tweeter has a 4" face plate which is not normal in a 3 way centre speaker. A tweeter with a lot smaller front is likely to be a better choice. This will enable a more efficient midrange to be an option.

That's a good point. I'll take a look at some other tweeters, but if I keep the D2905/950000 I figure that I'll get a new face plate made for the tweeter, I can't chop an awful lot off of it because the housing behind it is relatively big, but I expect I could reduce it to 3" in the vertical direction. I have an old D2905/9000 kicking around which is the same housing as the D2905/950000, so I'll take a look tonight if I remember. If I want the same sound then I guess I am best going for an MTM design, but with putting the mid/bass units as close to each other as possible and mounting the tweeter centrally above them? The thing that I'm not sure about when considering phase, is that since the whole cone moves, it's not exactly a point source, in which case, am I safe assuming that I can measure the distance from the edge of one mid/bass cone to the edge of the other mid/bass cone (in order to calculate the frequency at which horizontal phase will be an issue), or should I treat the cone as if it is a point source, and therefore measure from the centre of one cone to the centre of the other? If the latter, then I think I'd rather do a W/M-T/W design knowing that the sound won't be quite the same (I'm sure it would be a lot closer than what I have though) but at the same time, I shouldn't have as big an interference problem (am I worrying about something that is more a theoretical problem, than a an issue that is practically detectable by my ears?).

There is the Audax HM100Z0 if you want a more efficient midrange. Another option is using the midrange as a filler driver with output from the woofer and tweeter making up for the lack of sensitivity in the midrange. Another is using a tweeter with a small face giving room for a larger midrange.

That's a nice suggestion, I notice that Falcon Acoustics have one left, so that means I'll not end up with an extra one sitting around doing nothing.

Baffle diffraction puts dips and peaks in the frequency response, the drivers have dips and peaks in their responses, the crossover puts dips and peaks in the response,... The better designs will juggle with all these to even out the frequency response but it may not always be apparent from looking at the finished speaker. If you look at the better commercial speaker designs these days there have clearly been efforts to reduce the diffraction from the woofers which is less of an option for DIYers.

That's a good point, I'll see I can solve some of it by getting the baffle shape, size and driver placement right before trying any corrections in the crossover. Are there any tools for baffle diffraction that you would recommend that can simulated the effect of odd shaped baffles (possibly with curves on them)?

 

[aamacgregor]

Gravesen has a number of good center channel designs you might like here.

I had a look at those a couple of weeks ago, on the one hand I know that going with a trusted design is the sensible thing to do, but on the other hand I was planning on trying to get a uniform sound from the front and centre channels though, which is why I wanted to use the same drivers.

Your plan is good, but it's also quite ambitious. To go this far from scratch you need to learn how to analyze a speaker system and doing that well needs tools and education. If you want to get something quickly that works well, find a design from Gravesen or others.

I'm more than happy to spend the time and effort to gain the right experience (i.e. that old cycle of some good guided learning of the right information, suffering the pain of the failure because I didn't know something and having to start again... and again... and again...). On the education front, I think I'm ok... well kind of, possibly, maybe... I have a BSC Hons in Instrumentation with Applied Physics (amoung a lot of other things it covered high pass and low pass filters, measurement techniques and stuff, although it didn't exactly focus on sound, it did include it) and a PhD in Experimental Particle Physics (ok, that had absolutely nothing to do with audio and really doesn't help me here, but if nothing else should say that I won't give up easily). I know I have a lot that I need to learn in order to properly design and build my centre speaker, but I have you guys to help direct me (I really value the input from you all - your experience and understanding on speaker design is more valuable to me than my education, even those people that have only ever built one speaker - that's still more than I have built!). To be honest, I find learning fun, and the challenge and satisfaction of designing something that I can use and appreciate is the thing that drives me.

You'll also need to analyze the driver's T/s parameters and the in-cabinet impedance curves. For that you'll need Dayton DATS V2 or similar.

If rather than using the 18W/8545-01 (the improved slightly more sensitive version), I use the 18W/8545-00 (the actual drivers in my main 2 way speakers); then as a short-cut, can I skip this step by measuring the internal volume of my main speakers and their port dimensions, and replicate them? In my case, since I plan on using two of the same driver, I assume I need to double the internal volume in the cabinet and two ports instead of one, but keep each port the same size?

Audiohobby.eu only sells in pairs, but Madisound sells singles. I think.
Lastly, because they are off-spec, getting good bass with the Scanspeak's requires a lot more volume than you would think. If you have it great, but otherwise your centers could be huge. SS is notorious in this line for being off spec. Look at Madisound's recommended cabinet sizes as a better guide than the published specs.

That's good to know. Much appreciated info there.

 

[aamacgregor]

Is your AVR happy with 4ohm nominal speakers? Running 2 x 8ohm woofers in parallel will give you 4ohm.

I looked up the specs and thankfully, it is happy with a 4ohm nominal load.

And what kind of placement are you looking at for the CC? If it sits on a small stand out in open space, it will see the normal 6dB of baffle step loss and will probably work as planned. If on the other hand, you are fitting it on a shelf in something like an entertainment center, then you may not really see much baffle step loss at all in which case 2 of the 8ohm woofers in parallel may be too high in SPL to match your mid. Even if you go with the 4ohm 10F4424.

Hmmm, that's a little concerning, is it a +3dB rise that I get from two woofers in parallel? If I remember correctly (which knowing me I probably don't) the 18W/8545-00 is slightly less sensitive than the 18W/8545-01, so perhaps the former is the better bet? I'm planning to set the centre speaker on the top shelf of my AV stand (the TV will be mounted on the wall above it). Height is a concern (the taller the speaker, the further the TV gets pushed up the wall), but I'm ok for depth and width. Ideally, I'd probably want to limit cabinet height to 9" or perhaps 10" max, but would be happier if it were closer to 8". The cabinet depth can be up to about 18" and the width about 18".

Before purchasing anything, you should model and simulate your design. This will give you a better idea about feasibility.

I'm not sure what simulator is best to use here. I know there are a lot kicking around of various ages. I downloaded XSim, but couldn't get the part number lookup feature to work (it didn't see any part numbers, for anything, basically the list was empty). What about BoxSim? I don't mind paying up to £50 or so for a simulator if there is one out there for around that price (or less) that is a lot better than the free ones?

Try Falcon Acoustics for your parts source.

I've used them before when fixing my main speakers when one of the tweeters failed. Nice people and good prices.

 

[eriksquires]

You can use your current speakers as a guide, but remember you'll need to multiply the volume by the number of drivers.

You know, I thought it was 3 dB as well, but XSim says 6 dB. rise from parallel drivers, but only if parallel. If you run them in series this goes away. I remember Dr. Leach a long time ago went through the effects on damping when you do this, turns out it's not that big a deal.



Best,

Erik

 

[jReave]

Yes, 2 x drivers in parallel: impedance is halved and SPL gain = +6dB
and, 2 x drivers in series: impedance is doubled and SPL gain = 0dB

So you might also want to consider the option of 2 of the 18W4545, but that's getting a little ahead of ourselves.

I asked about placement because it affects the woofer SPL's and this level is going to determine what the minimal SPL requirements will be for the mid and the tweeter too. So sitting on top of an AV stand might be better than inside one but it still might help to know more about the stand because when you sit the speaker on it, the dimensions of the stand more or less act like an extension of the front baffle. More so if the front of the stand is one solid surface (with doors or drawers or glass), less so if it is more open. Maybe post a picture or diagram of it and then we can see what the diffraction might look like for the whole thing. For that you'll want to use The Edge diffraction simulator.

XSim btw is an excellent xo program but it doesn't have any drivers in its data base yet and probably never will. You have to input your driver FR and impedance files into it (left click on the driver and chose "Tune") the same way that you do for the other popular programs, PCD and WinPCD. I do particularly recommend PCD as it's got a couple of extra features in it that can be important to a CC.

For more info into the simulation process try The Speaker Building Bible, starting at the xo section. I'm currently trying to help someone else with this process; you may find some of that helpful as well if for nothing more than some of the links:
xsim-crossover-critique
xsim-critique-part-2

 

[andy19191]

Are there any tools for baffle diffraction that you would recommend that can simulated the effect of odd shaped baffles (possibly with curves on them)?

The accurate handling of curves would require a 3D simulation (e.g. BEM). Some of the simple "2D"(ish) baffle diffraction simulators can make approximations for chamfers or radiused edges. If you place the speakers on a shelf with other surfaces close to the baffle these will also have a significant influence. Is there clean air above the tweeter or a shelf with stuff on it?

The diagrams here might be informative. Some centre speaker examples further down the page.

Jeff bagby's Excel spreadsheets are a fairly complete speaker design suite that is quite widely used. It includes baffle edge diffraction, room boundary reinforcement along with the passive crossover design enabling you to fiddle with all effects in the same place. Unfortunately I have never used the software because I have never bought a copy of Excel (OpenOffice does not work). There is a project WinPCD to get the models out of Excel and into something less restrictive but I don't know how complete it is. I downloaded the program but the setup failed with a 404 error. If you ignore this and click on the WinPCD.exe file it seems to run. I already have the .Net bloat installed and I think the setup failure is unimportant and related to this.

As other have said you get a 6dB rise in efficiency in parallel: +3dB from twice the area and +3dB from half the resistance drawing twice as much current. In series you get +3dB from twice the area but -3dB from twice the resistance drawing half as much current.

The simplest option is to squeeze the midwoofers as close together as possible, cut two arcs out of the tweeter and put it as close as possible above and between the midwoofers. On axis this should be reasonable but off to the side the tweeter and midwoofer will start to cancel. To check how much of a problem it might be you can lay your current speaker on it's side and sit on- and off-axis. The tweeter centre to midwoofer centre in this case will be larger than the proposed centre speaker design and so is a conservative test.

 

[aamacgregor]

You can use your current speakers as a guide, but remember you'll need to multiply the volume by the number of drivers.

You know, I thought it was 3 dB as well, but XSim says 6 dB. rise from parallel drivers, but only if parallel. If you run them in series this goes away. I remember Dr. Leach a long time ago went through the effects on damping when you do this, turns out it's not that big a deal.



Best,

Erik

Thanks Erik, I think you are right, I'll need to run them in parallel, I'm beginning to think that getting two of the 4 ohm versions as suggested by jReave is the thing to do.

 

[aamacgregor]

Yes, 2 x drivers in parallel: impedance is halved and SPL gain = +6dB
and, 2 x drivers in series: impedance is doubled and SPL gain = 0dB

So you might also want to consider the option of 2 of the 18W4545, but that's getting a little ahead of ourselves.

You know what, I think your suggestion that I consider the 4ohm ones is probably the right way to go. In fact it is the only way that I can go as far as I see, since if I get the 8ohm versions then I either end up with a 16ohm load, or an SPL that is far to high to match against a mid that is in the size range that I'm after.

I asked about placement because it affects the woofer SPL's and this level is going to determine what the minimal SPL requirements will be for the mid and the tweeter too. So sitting on top of an AV stand might be better than inside one but it still might help to know more about the stand because when you sit the speaker on it, the dimensions of the stand more or less act like an extension of the front baffle. More so if the front of the stand is one solid surface (with doors or drawers or glass), less so if it is more open. Maybe post a picture or diagram of it and then we can see what the diffraction might look like for the whole thing. For that you'll want to use The Edge diffraction simulator.

So, here is the thing, before I can build the centre speaker, I need to build (or get one built) a new AV stand. At the moment, I have a three shelf AV stand; the TV sits on the top and takes up the whole top (it's an old 40" Sony Bravia LCD screen with a thick boarder on it); the middle shelf which is probably only 15cm in height has the centre speaker on it and behind that on the same shelf is the external crossover for my main speakers along with an 8 port Netgear switch; the bottom shelf has a my server that sits in a full size HTPC case (Zalman HD160 Plus). Then next to that I have a 4 shelf Stands Unique hifi stand that has my hifi equipment and AV receiver on it. The thing is, there are a lot of cables that are visible between the stands and down the back of the Stands Unique hifi stand, and well... my wife doesn't like to see the cables. Plus I plan on changing the TV to a 55" TV and the AV stand won't take a TV that size. So my plan is to design a 3 component wide, 3 shelf AV stand with a TV support built on the back of it that will support my planned TV. As far as dimensions go, I think the stand will be about 1.75m wide (~5' 9") and 65cm deep (2' 2"). The reason why it will be so deep is because I have a McIntosh MA6300 that powers my main speakers that its specs say is 22" (~56cm) deep and I need to leave cable room since I plan on putting a solid back on the stand (but leave the sides open with pillar supports near the front corners to support between the shelves at the front). You know what, my description is hard to visualise. I'll draw a diagram and upload it later.

XSim btw is an excellent xo program but it doesn't have any drivers in its data base yet and probably never will. You have to input your driver FR and impedance files into it (left click on the driver and chose "Tune") the same way that you do for the other popular programs, PCD and WinPCD. I do particularly recommend PCD as it's got a couple of extra features in it that can be important to a CC.

Thank you so much for the links, I didn't know about Jeff Babgy let alone his Excel spreadsheets, which are amazing (I downloaded them last night and had a quick play - which told me that I'm getting massive boundary effects from my main speakers at the moment because they are so close to both the back and side walls!). Also, thanks for letting me know there are no drivers in the driver database for Xsim. I managed to find some FRD and ZMA files for the ScanSpeak D2905/950000, but haven't found any for the 18W/8545 (or more to the point the 18W/4545). I didn't look to hard to be fair though, but if I can't find any then I guess I'll need to plot them myself.

For more info into the simulation process try The Speaker Building Bible, starting at the xo section. I'm currently trying to help someone else with this process; you may find some of that helpful as well if for nothing more than some of the links:
xsim-crossover-critique
xsim-critique-part-2

Those links are amazing! Thank you so much. Part of the reason why I've taken so long to reply is because I can't stop myself from reading from the links that you supplied (although the main reason is that I haven't had the opportunity to write back using my PC - trying to reply on a phone is time consuming and frustrating).

 

[aamacgregor]

The accurate handling of curves would require a 3D simulation (e.g. BEM). Some of the simple "2D"(ish) baffle diffraction simulators can make approximations for chamfers or radiused edges. If you place the speakers on a shelf with other surfaces close to the baffle these will also have a significant influence. Is there clean air above the tweeter or a shelf with stuff on it?

Well, with the new stand that I'm designing, I plan on putting the centre speaker on the top shelf, the record deck would probably sit relatively close to it on one side (unless I mount it on the wall, which is an option), and the CD player would be on the other side (although if I mount the record deck on the wall, then I'd put the CD player on the middle shelf, in which case, there would be nothing either side of the centre speaker)

The diagrams here might be informative. Some centre speaker examples further down the page.

Thanks for the link, I really appreciate it, there are some nice designs on there. You know, as I've been learning over the last few days, I've started looking at different speakers and judging them with prejudice based on their visible design. For example, I saw this centre speaker on another page that had a M-T-M-M-T-M configuration where all the drivers are aligned on the same plane horizontally - my immediate thought was "that must have terrible interference issues with the drivers spaced like that!". Last week if I'd seen a picture like that, I might have simply thought "fair enough".

Jeff bagby's Excel spreadsheets are a fairly complete speaker design suite that is quite widely used. It includes baffle edge diffraction, room boundary reinforcement along with the passive crossover design enabling you to fiddle with all effects in the same place. Unfortunately I have never used the software because I have never bought a copy of Excel (OpenOffice does not work). There is a project WinPCD to get the models out of Excel and into something less restrictive but I don't know how complete it is. I downloaded the program but the setup failed with a 404 error. If you ignore this and click on the WinPCD.exe file it seems to run. I already have the .Net bloat installed and I think the setup failure is unimportant and related to this.

I didn't know that WinPCD might be able to import from Jeff's spread sheets, that's pretty neat if it works. I haven't downloaded WinPCD yet, but I have taken a look at the spread sheets and they look great.

As other have said you get a 6dB rise in efficiency in parallel: +3dB from twice the area and +3dB from half the resistance drawing twice as much current. In series you get +3dB from twice the area but -3dB from twice the resistance drawing half as much current.

Oh so that's where the +6dB comes from. That is a really helpful explanation - thank you very much.

The simplest option is to squeeze the midwoofers as close together as possible, cut two arcs out of the tweeter and put it as close as possible above and between the midwoofers. On axis this should be reasonable but off to the side the tweeter and midwoofer will start to cancel. To check how much of a problem it might be you can lay your current speaker on it's side and sit on- and off-axis. The tweeter centre to midwoofer centre in this case will be larger than the proposed centre speaker design and so is a conservative test.

That's a pretty good idea [puts one speaker on it's side and disconnects the other for a test]. Ok, so I know the mid/bass unit is further away from the tweeter than you are suggesting for the centre speaker, so if I could get the mid/bass units close together (so that they are almost touching) and then cut the arcs out of the tweeter face place as you suggest in order to not push the tweeter to high on the baffle, then the interference should be reduced to the point that it may no longer be detectable; but from a quick test, I'm concerned about what the off axis response will be like if I don't go 3 way even if I can get the two mid/bass drivers really close together. My main concern here is I can definitely detect beaming from the mid/bass and I "think" I can detect a slight phasing issue (interference peaks and troughs) somewhere in the crossover region at the upper end of the mid/bass FR (it's more noticeable if I stick my finger in one ear and move my head from side to side on certain tracks). I'm guessing that means that by the time interference starts to be a problem that the tweeter is at a higher SPL than the mid SPL at that frequency. I might do a proper measured test at various frequencies using a signal generator app and microphone at my listening position (moving the mix from left to right while a constant tone is playing) since just playing music isn't really a good test - especially I might suddenly come across a piece of music where it's really noticeable. To be honest, my bigger concern at the moment is the beaming in the upper mid from the mid/bass driver. If I use a small 4" mid, then the beaming effect will be reduced right?

Here is another thing, someone suggested to me, that part of the reason (they suggested that it might possibly be more significant than I might think) why my current centre speaker sounds compressed by comparison to my mains could be to do with the fact that my mains are being powered by a much better quality amp - I have a Pioneer 2023-K powering the centre and two surround channels and am using the front pre-out into a line-in on my McIntosh MA6300 to power the fronts. Although I take that point, I'm pretty convinced that the centre speaker itself is the biggest factor. Either way, I want to build a centre speaker, but I might consider getting a separate power amp to power the centre and rear channels at some point just to test that theory.

 

[aamacgregor]

By the way, some useful information might be that my listening room is 2.87m (9'5") wide (the TV, front and centre channels are on this wall) and 3.8m (12' 6") long. My sofa is against the back wall, so I guess my head will be about 2.6m (8' 6") from the centre channel.

 

[eriksquires]

This is all still looking quite complicated. I haven't really read where things had gotten to. May I suggest a 2.5 way design, and measure your drivers in far field and in place, instead of attempting quasi-anechoic measurements/simulations/blah blah blah?

The beauty of a 2.5 way here is that you minimize the complicated phase interactions to really just the tweeter and mid-woofer, while the twin woofer stays out of it. Also, you end up with almost the same dB of boost from the dual radiators. Anywhere from 4 to 6 dB, depending.

To be clear, what I mean is you use identical woofers, one center located tweeter.

One woofer acts as the mid-woofer, the other the pure, or true woofer.

Use as high an order of filters between the mid-woofer and the tweeter as you can get. LR4 ideal, but not always easy. Cross them over otherwise normally, around 2kHz depending.

Use no low pass on the mid-woofer.

Use low pass filter on the woofer, around 400-200Hz as needed to support the mid-woofer and minimize BSC needed. 2nd order should be fine. The lower you go the more you can ignore phase matching issues.

The woofer is cut off so low (200 - 400 Hz) that the frequencies are quite long and less likely to significantly interact with it's twin the mid-woofer, and has no interaction with the tweeter. have some one spin the speaker around randomly and you won't even be able to tell which is which from 9 feet away.

You can experiment with this rather quickly with XSim. Try some ideal drivers, and ideal crossovers and play with the distances. For a 6" woofer and tweeter, I imagine around 1.3" is going to be really close to the true distance. Try setting the mid-woofer to zero, and set the tweeter to 1.3" and woofer to 2", and vice versa. I think you'll find it's a decent answer to the center-channel MTM problem.

This is what is referred to as the classic Squires Center Alignment (SCA).

Best,


Erik

 

[andy19191]

Ok, so I know the mid/bass unit is further away from the tweeter than you are suggesting for the centre speaker, so if I could get the mid/bass units close together (so that they are almost touching) and then cut the arcs out of the tweeter face place as you suggest in order to not push the tweeter to high on the baffle, then the interference should be reduced to the point that it may no longer be detectable;

Not really. All high performance centre speakers will have a tweeter above a midrange or a coaxial. What was being suggested would be cheaper and simpler than a 3 way and preserve the centre speaker sounding like your main speakers to the maximum extent. If you add a midrange your centre speaker will sound significantly different to your main speaker and so there is no point using a tweeter with a 4" face plate in a 3 way centre design and damaging the choice of midrange which is probably the most important driver of the three in terms of sound quality.

If I use a small 4" mid, then the beaming effect will be reduced right?

It will noticeably smooth the sharp change in directivity from midwoofer to tweeter in your main speakers. It will start beaming at a higher frequency than the midwoofer in your mains but I would need to check the degree of significance.

Here is another thing, someone suggested to me, that part of the reason (they suggested that it might possibly be more significant than I might think) why my current centre speaker sounds compressed by comparison to my mains could be to do with the fact that my mains are being powered by a much better quality amp

As a sweeping generalisation, reasonable hi-fi amplifiers operating within their limits tend to be audibly neutral but some of the cheaper AV receivers/amplifiers can have issues. I don't know in your case but I would not dismiss the possibility.

 

[jReave]

I think Andy and Eric are both right about a WTW center being the most similar to your mains. Going 2.5-way as Eric suggests would also be the smart thing to do in this case as it would avoid most of (if not all) of the lobing running both woofers full out would produce.

However, I would expect the 3-way to be a better sounding speaker so for me it would sort of become a question of uniformity vs quality. Mind you, you do like your current mains so those drivers may in fact be just fine for the CC. On the other hand, although it's generally recommended, I only really notice a non-uniformity issue when there is an obvious panning of action across speakers and I've found that this happens so infrequently in movies that it doesn't bother me if the CC isn't exactly the same as the other speakers in these kinds of situations. My biggest peave with CC's is dialogue intelligibility and for that I want the best quality I can get, which for me is with a dedicated mid relieved of LF duties sitting directly under the tweeter. Just my 2 cents anyways.

I'm not sure if you can run your current CC on your Macintosh but you should certainly be able to run your mains on the Pioneer. That should give you some insight into whether it's your CC amp that's the problem or not.

I've attached a rough sim of a CC on top of or inside your planned AV stand when the stands front face acts as an extension of the baffle which won't be completely true but close enough for my purposes here. It does end up seeing almost the whole 6dB of baffle step loss, but it happens at much lower frequencies than usual. When you add these effects to your woofer, the approximate 2dB loss (ie. 2 dB down from 6dB) at about 200Hz and 250Hz is about the amount of loss that the driver will end up seeing when you add in the xo to correct for the gain that's above those levels. Let me know if that's not clear.

So if you use 2 x 18W/8545 (@ about 87dB sensitivity) in parallel, you'll end up with a speaker at about 91dB (6dB gain - 2dB loss). You actually have a few high quality 90dB or so mids to choose from that should be able to work here, like:
Audax HM100Z0 Aerogel 10cm 4" woofer
Peerless Vifa NE123W-04 MidWoofer Speaker
Scanspeak 10F/4424G00 Mid Range - Discovery Range
Your problem actually isn't with the mid but with your current tweeter which is only 90dB sensitivity. In this case you would need to choose a different and perhaps preferably smaller faceplate tweeter for a good match.

With 2 x 18W4545 @ about 89dB in series, you'll end up with about 87dB sensitivity after diffraction. So more mid possibilities here and your current tweeter would still work fine.

In both cases, you are likely to find that a vented alignment won't really help very much because the majority of the diffraction loss happens primarily below 100Hz which is where you'd get the gain from the vented enclosure. So no gain at all in other words with the vented alignment.

 

[eriksquires]

JWeave,

Yes, 3 way is better, I usually find it rather madening to try to find a quality tweeter and midrange that will fit vertically in the typically low profile of a center channel speaker. Much easier in a standalone that isn't being squeezed under an HD television.

But hey, if you have the space, do it!

Here's an example I'm working on for my blog. It uses much cheaper parts, but the phase alignment issues are the same. I've attached the XSim data so you can play with it to see what I'm talking about regarding how the phase issues go away, and how the extra driver takes care of the BSC for you. This is a work in progress so the part values have not been normalized nor checked for power ratings, etc. I'm only sharing this to help illustrate the issues being discussed.

Regardless of which driver you set as your 0 point, the FR doesn't change very much.

Best,

Erik

 

[eriksquires]

Here's the FR, with individual driver plots. Notes:

• The target curve I'm using is my own. Don't mind if you don't like it.
• The woofer measurements are in far field in a bookshelf. Deliberately!

Again, you don't have to like my choices, the point here is about phase problems with a WTW setup.



Best,




Erik

 

[aamacgregor]

This is all still looking quite complicated.

Yeah, and I think it's about to get worse.

I like really like your 2.5 way idea and it's something that I'm certainly considering as the cheaper option that will probably do what I want. I'm also wondering if the beaming issue that I was concerned about is less of an issue than I made out - basically, I was sitting on the left side of the room, but I turned the right speaker on its side and pointed it so that driver was pointing straight at the back wall (i.e. no tow in) in order to simulate an over done off-axis response from the centre speaker, but I think I was a bit unfair there. If I was going to face the speaker straight at the back wall, I should have at least placed it at the intended centre position. Anyway, I should probably run that test again before discarding the 2.5 way idea.