Active center speaker on a budget

[voodooless]

My very first serious DIY speaker project was a centre speaker about two decades ago. Back then I was already aware of the fact that a horizontal center speaker usually does not have a decent off-axis response. So my center was vertical, comprising of 2x Monacor SPH-135KEP, one of the lower distortion woofers, and a ribbon tweeter (don't remember the model). For a first attempt, it worked pretty well

Well, a lot of time passed since those days, and I've been using a stereo system for basically all that time. But I actually want to move towards a multichannel setup, and the center is the first thing on the list. This brings us back to today I'd like to make another center, but this time the requirements are a bit different:

• Since the unit will be placed underneath my TV on a cabinet, it can only be 11 cm in height. I might stretch it to 12, but that is really all I have to offer In the horizontal direction, I have a lot more room, up to 80 cm would be no problem. Depth: 30 cm maximum.
• Good on and off-axis response.
• It needs to be active and DSP powered.
• All components need to be in the box. I don't have a lot of room for all kinds of amp and DSP boxes.
• It needs to be able to play fairly loud, let's say above 103 dB at 1m, especially at the lower end
• I don't need maximum bass extension, so 80 Hz crossover would be fine. If it goes lower, I'll take it. I don't mind reflex. It'll probably be needed to get to the SPL target anyway
• I'd like to keep the budget relatively low. Let's say € 500,- for the whole thing.

So here is my take on this, and I'd like you guys to comment on it . So we have only 11 cm of height, which means that we cannot fit a tweeter and woofer vertically. So a small coaxial is needed. Usually, these things will not go very loud, but I found one exception: B&C 4CXN36. A 4" PA coax that actually looks pretty good overall. Next, we'll need some serious output below something like 300 Hz. Tried several 4" woofers and ended up with a very budget-friendly option: LaVoce WSF041.00. Four of them should give enough output to make me happy. In a reflex enclosure, this should also dig down to about 60 Hz.

Now for the DSP part, there are not very many options on a budget. Basically, the only option would be something with ADAU1701. It does not have the best DAC and ADC, but it's probably good enough, and most boards offer at least digital in- and output that might yield future DAC or ADC upgrades. My current best option is the 3e audio one, which was also tested by Amir. Next, I'll need some amps to drive all of this. Luckily 3e audio also has a 4 channel 100W module available that fits the DSP module very well. All of this needs a power supply. I opted for a 300W supply found on Aliexpress. It also provides the aux needed for the DSP and is fanless. Other options would be nice, however, a fanless supply with more than 250W is hard to find. Finally, we'll need to program the DSP. Since I have a Mac, I can only run Sigmastudio via Crossover, and that means any USB option basically is non-functional. I don't like the cable options anyway. So my idea would be to use a cheap ESP32 as a bridge so I can access the DSP over WiFi.

Made a small "infographic" depicting all the components.

The next steps will be to gather some more input, do some more sims, and see if it all fits into a nice enclosure. Then it's time to order some stuff.

 

[alex-z]

Looking at the 4CXN36, it has good power handling and sensitivity, but the directivity is going to hurt. Even if you cross at 2000Hz to make them match initially, the compression driver narrows from 4000-8000Hz.

I think it would make more sense to just use a convential tweeter and mid-range, placed horizontally as close as possible. If you can get the centre to centre spacing low enough it will work. Something like an SB10PGC214 + Dayton RST28-A gets you to 95mm C-C, and a 2000Hz crossover would eliminate directivity compromise.

Woofer choice looks fine, although if you can stretch the height to 12cm then the Peerless SDS-P830855 or Dayton RS125T have lower Fs, and are 105 and 103mm respectively. Cabinet walls only being ~10mm thick isn't great, but managable if braced heavily.

Doing compact + DSP, I would definitely favour a sealed design. Doing 103dB at 1 metre means the port would take up valuable space, possibly an entire drivers worth.

Looking at the SDS-P830855 specifically, Linkwitz Transform down to an F3 of 70Hz + high-pass at 70Hz puts the power handling at 5.5 watts per driver, 91.5dB at 1 metre, +12dB for 4 drivers in parallel meets your 103dB criteria.

Edit: How is the centre channel being placed exactly? If the top or bottom had some clearance, you could do an 8-10" passive radiator in place of a port.

 

[gy-k]

For a power supply there's the Meanwell LRS-200, with 24, 36, 48V variants. If you want to skimp on amplifier power there are cheap TPA3116 based 6-channel amp boards that have 4x BTL and 2x PBTL channels.

Sica have a new 4" coaxial driver, model 4C1,5CP that has a 1.5" voice coil and a headline sensitivity rating of 91dB. So looks like higher sensitivity rating than for the 4" B&C cone, but with a dome tweeter, though it's probably not available anywhere to buy currently. But the B&C 4CXN36 looks pretty cool, I can imagine the HF unit could be used down to 1300Hz.

The LaVoce WSF041.00 driver has cutout diameter of 93.5mm, adding 2x 10mm for the walls of the box, I think we're looking at a height of 11.35cm minimum. (Alternatively there are 4" drivers from Peerless with truncated frames with similar dimensions.)

Alternative to coax driver would be 3D printing a small Synergy horn thingy. Like generating a horn that fits the box with ATH4 and adding elements to the 3D model to which woofers on two sides of the horn could be mounted. 2x Sica 3,5H would fit inside the box, giving a bit more cone area. A small 60EUR horn driver like the BMS 4524 could probably play down to 1200Hz. It wouldn't be cheaper overall, but I guess something that works could be done even without getting too deep into the subject, just by having the slots for the woofers that have the right size "intuitively" even, and observing the 1/4 wavelength distance rule, especially there's DSP to shape the responses of the drivers.

 

[voodooless]

Thanks guys, appreciate the comments!

Looking at the 4CXN36, it has good power handling and sensitivity, but the directivity is going to hurt. Even if you cross at 2000Hz to make them match initially, the compression driver narrows from 4000-8000Hz.

It’s really not that bad looking at the polars, is it? A 25mm dome would also narrow. Just look at the Sica Coax polars.

I think it would make more sense to just use a convential tweeter and mid-range, placed horizontally as close as possible.

I doubt that would give proper polars, even if you can place them very close

If you can get the centre to centre spacing low enough it will work. Something like an SB10PGC214 + Dayton RST28-A gets you to 95mm C-C, and a 2000Hz crossover would eliminate directivity compromise.

Nah, it never does. Nor will the SB hold up at high volume.

Woofer choice looks fine, although if you can stretch the height to 12cm then the Peerless SDS-P830855 or Dayton RS125T have lower Fs, and are 105 and 103mm respectively. Cabinet walls only being ~10mm thick isn't great, but managable if braced heavily.

They don’t have to be only 10 mm thick. Some clever woodworking will take care of that. 12 cm sadly does not fit. It would have given a lot more options indeed.

Doing compact + DSP, I would definitely favour a sealed design. Doing 103dB at 1 metre means the port would take up valuable space, possibly an entire drivers worth.

Looking at the SDS-P830855 specifically, Linkwitz Transform down to an F3 of 70Hz + high-pass at 70Hz puts the power handling at 5.5 watts per driver, 91.5dB at 1 metre, +12dB for 4 drivers in parallel meets your 103dB criteria.

I’ll do some more sims. To work closed down to 80 Hz, a lower FS would make a world of difference.

Edit: How is the centre channel being placed exactly? If the top or bottom had some clearance, you could do an 8-10" passive radiator in place of a port.

I’ve been thinking of this. The top would be partially obscured by the TV, but might fit a PR. Top lotaction is however far from ideal for a PR because it sags and will be in a asymmetric state by default. Anyone experience with this? Alternatively, PR’s on the sides or back? Also note that this will add cost.

For a power supply there's the Meanwell LRS-200, with 24, 36, 48V variants.

I’ve seen that one. It was on the list for a while. However it’s only 200W and is missing the AUX. higher power version has a fan.

If you want to skimp on amplifier power there are cheap TPA3116 based 6-channel amp boards that have 4x BTL and 2x PBTL channels.

Not sure that will be a better match or will lower the budget?

Sica have a new 4" coaxial driver, model 4C1,5CP that has a 1.5" voice coil and a headline sensitivity rating of 91dB. So looks like higher sensitivity rating than for the 4" B&C cone, but with a dome tweeter, though it's probably not available anywhere to buy currently. But the B&C 4CXN36 looks pretty cool, I can imagine the HF unit could be used down to 1300Hz.

Have not found that Sica yet, thanks! Looks pretty nice overall. Indeed hard to buy at the moment.

The LaVoce WSF041.00 driver has cutout diameter of 93.5mm, adding 2x 10mm for the walls of the box, I think we're looking at a height of 11.35cm minimum. (Alternatively there are 4" drivers from Peerless with truncated frames with similar dimensions.)

As said before, we can sneak in some additional mm if needed

Alternative to coax driver would be 3D printing a small Synergy horn thingy. Like generating a horn that fits the box with ATH4 and adding elements to the 3D model to which woofers on two sides of the horn could be mounted. 2x Sica 3,5H would fit inside the box, giving a bit more cone area. A small 60EUR horn driver like the BMS 4524 could probably play down to 1200Hz. It wouldn't be cheaper overall, but I guess something that works could be done even without getting too deep into the subject, just by having the slots for the woofers that have the right size "intuitively" even, and observing the 1/4 wavelength distance rule, especially there's DSP to shape the responses of the drivers.

That is a cool idea. However I think the difference in angle between horizontal and vertical will be tricky to get right. The vertical will probably not be able to hold pattern long low enough. Also the vertical angle might be to narrow.

 

[ppataki]

Have you considered going with fullrange drivers, no crossover?
If you will have DSP anyway, you could just equalize the response accordingly
Maybe 5-6 pieces of ND91-8 in a Bessel array? Or 4-6-8 of them in a series-parallel connected array?
Just thinking out loud here...

 

[voodooless]

Have you considered going with fullrange drivers, no crossover?
If you will have DSP anyway, you could just equalize the response accordingly

Yes! I have a whole topic about it .

Maybe 5-6 pieces of ND91-8 in a Bessel array? Or 4-6-8 of them in a series-parallel connected array?
Just thinking out loud here...

I would have considered it an option if the listing position would be further away. A normal array will just have terrible polars, so won’t do either. A CBT array might be better, but I’d rather have a flat baffle.

 

[ppataki]

If you go with normal line array and not Bessel with the ND91-8 (like 4-6-8 of those) I would not think that large distance would be that critical but I might be wrong

I know that post, you remember we discussed the Bessel topic there that my friend is venturing into with the 5 pieces of CHR120
I will report back once that is finished, in a few months time hopefully

 

[voodooless]

If you go with normal line array and not Bessel with the ND91-8 (like 4-6-8 of those) I would not think that large distance would be that critical but I might be wrong

Distance is not a problem, directivity is. It’s just the wrong way around

 

[somebodyelse]

Does SigmaStudio work in Crossover? I tried with Wine in linux, but IIRC it wouldn't start because it couldn't load the USB driver, even though it's not strictly necessary as programming over the network is an option. Perhaps I should have tried harder, instead of using the time limited Windows VM developer images from MS.

 

[voodooless]

Does SigmaStudio work in Crossover? I tried with Wine in linux, but IIRC it wouldn't start because it couldn't load the USB driver, even though it's not strictly necessary as programming over the network is an option. Perhaps I should have tried harder, instead of using the time limited Windows VM developer images from MS.

Yes it does. Used it successfully on a HifiBerry DSP board.

 

[alex-z]

Thanks guys, appreciate the comments!


It’s really not that bad looking at the polars, is it? A 25mm dome would also narrow. Just look at the Sica Coax polars.

I doubt that would give proper polars, even if you can place them very close

Nah, it never does. Nor will the SB hold up at high volume.

They don’t have to be only 10 mm thick. Some clever woodworking will take care of that. 12 cm sadly does not fit. It would have given a lot more options indeed.

Yes, a dome tweeter would narrow, but unlike the coaxial it wouldn't get 15-20 degrees wider again above 8000Hz. The PA style of coaxial drivers are just far inferior to the variety with mounted dome tweeters, especially the KEF or Genelec design that actually optimizes for directivity. Unfortunately you can't use a salvaged Q150 with the ~11cm height limitation, so optimized 3 way design is the next best thing.

You could get good polar response with a 2000Hz crossover. It won't be perfect, but the narrowing will be minor, only 3-5dB for ~500Hz window at the crossover region. Possibly even less if you removed the stock RST28-A faceplate and created a mild waveguide.

103dB might be a stretch for the SB10PGC, but I think it would work better than you expect. The 20 watt rated power rating is 200-20000Hz pink noise with a 12dB high-pass. Raising the crossover to 400Hz and 24dB should give you considerably more power handling.

 

[voodooless]

Yes, a dome tweeter would narrow, but unlike the coaxial it wouldn't get 15-20 degrees wider again above 8000Hz. The PA style of coaxial drivers are just far inferior to the variety with mounted dome tweeters, especially the KEF or Genelec design that actually optimizes for directivity. Unfortunately you can't use a salvaged Q150 with the ~11cm height limitation, so optimized 3 way design is the next best thing.

Sure the B&C is not the ideal coax. Nothing beats KEF at their game at the moment. I would love to have a 10 cm KEF coax on hand. Having said that, I still think a coax will do better than a normal 2-way. Actually, the polar of the B&C looks very similar to the original LS50:

Having seen this, I doubt a Q150 unit would do much better.

You could get good polar response with a 2000Hz crossover. It won't be perfect, but the narrowing will be minor, only 3-5dB for ~500Hz window at the crossover region. Possibly even less if you removed the stock RST28-A faceplate and created a mild waveguide.

The question is if this is more acceptable than directivity errors higher up?

103dB might be a stretch for the SB10PGC, but I think it would work better than you expect. The 20 watt rated power rating is 200-20000Hz pink noise with a 12dB high-pass. Raising the crossover to 400Hz and 24dB should give you considerably more power handling.

It won't make anywhere near 103 dB:

This is MaxSPL: blue is the B&C coax, green is the SB. It's about 10 dB short! I would be better off with a 5th LaVoce driver in this case.

 

[voodooless]

This is the combined polar plot of the B&C coax. I photoshopped the crossover somewhere at 1.5 Khz. All this is without any EQ on the frequeny response, so end result should be a bit better.

 

[markymiles]

Will watch this with interest and see where you end up as I am in a similar position for the centre, and would probably make matching vertical LR too. I was considering the Beyma 5CX200ND as they seem well regarded, I take it you can't squeeze an extra 2-3 cm height into your design and going up a size in the drivers?

 

[somebodyelse]

I guess the LSX driver is a fraction too big at a nominal 4.5"
https://setelec-shop.fr/haut-parleur-kef/4298-haut-parleur-kef-lsx-serie-argent.html

 

[voodooless]

I guess the LSX driver is a fraction too big at a nominal 4.5"
https://setelec-shop.fr/haut-parleur-kef/4298-haut-parleur-kef-lsx-serie-argent.html

Are any size measurements available? 4.5" is probably stretching it, but the driver build is unusual, so why knows..

Although looking at the distortion plots, I don't think it will make it until even 100 dB:

Edit: did a rough estimate of the size, and it looks like it's 11.2 cm, possibly 11.5 cm. That would leave absolutely no wiggle room whatsoever.

 

[somebodyelse]

No further details - just thought it worth mentioning as it's smaller (I think...) than the Q150 mentioned earlier, and seemed borderline on size since the mount is narrower than on the Q150 and LS50 drivers. Erin's are the only measurements I've seen for bare KEF coax drivers, and they don't include the LSX one. Looks like the distortion goes through the roof at ~300Hz though.

 

[gy-k]

I’ve seen that one. It was on the list for a while. However it’s only 200W and is missing the AUX. higher power version has a fan.

I guess simulations will provide a good estimation about amp power needed to hit the Xmax of the drivers.

Have not found that Sica yet, thanks! Looks pretty nice overall. Indeed hard to buy at the moment.

Yeah didn't mean to complain about the B&C coaxial. It would be interesting to see how low the HF driver can go. Did a search for the HF diaphragm part number, it's the same as for the DE110-16 compression driver. There's a test of the DE110-8 here: https://audioxpress.com/article/test-bench-b-c-speakers-de110-8-compression-driver
Based on the data here I don't see anything why this one couldn't be crossed as low as 1300-1500Hz?

That is a cool idea. However I think the difference in angle between horizontal and vertical will be tricky to get right. The vertical will probably not be able to hold pattern long low enough. Also the vertical angle might be to narrow.

I just assumed it would be possible to generate a square shaped horn that has the same horizontal and vertical beamwidth and is also similarly wide like the cone of coaxial driver.

 

[voodooless]

I guess simulations will provide a good estimation about amp power needed to hit the Xmax of the drivers.

I did, it pretty much comes down to the 100W per channel. The CD less, but it’s 16 Ohm anyway.

Did a search for the HF diaphragm part number, it's the same as for the DE110-16 compression driver. There's a test of the DE110-8 here: https://audioxpress.com/article/test-bench-b-c-speakers-de110-8-compression-driver
Based on the data here I don't see anything why this one couldn't be crossed as low as 1300-1500Hz?

That looks very clean indeed! 1.3 kHz is probably possible.

I just assumed it would be possible to generate a square shaped horn that has the same horizontal and vertical beamwidth and is also similarly wide like the cone of coaxial driver.

Even the small ME10 in the audioxpress review is already 13cm. One could chop off the mounting plate to get down to size. One would still be very hard pressed getting any kind of midranges on there. Possibly 2x 3.5” might fit.

 

[voodooless]

I’ll do some more sims. To work closed down to 80 Hz, a lower FS would make a world of difference.

Nope, this is near impossible with 4x 4”. You’ll need almost 5 mm of Xmax to get to 103 dB. If these drivers exists, they will be above budget, and or have not enough power handling. You just can’t beat physics.