Help me to design a stereo pair of 3-way speakers from scratch (well, almost from scratch)

[Momomo67890]

Sica coaxial is vast superior to pretty much any other consumer coax available.

 

[Old Hi-Fi Guy]

Sica coaxial is vast superior to pretty much any other consumer coax available.

Thanks! The 5.5" looks promising.

 

[Old Hi-Fi Guy]

Wave guides.

I've looked at augerpro's. I think I'd need a 5", but it's available only in elliptical instead of circular. I'm a little confused about the positioning of a WG with respect to the midrange. If I understand correctly, the TM centre-to-centre distance is usually kept as small as possible, but I've also read that this distance should be 1.2 to 1.4 times the wavelength of the XO frequency. Trying to do both will push the XO frequency up quite high. Am I missing something? An elliptical WG seems to support keeping the distance small, but I've seen designs with circular WGs and quite large separations.

This got me thinking about other ways of doing wave guides. . .

I found a Peerless H26TG45-06 and a SB26STWGC-4 with integrated WGs. Are these intended for general use or for specific applications?

I suppose a wave guide has an exponential shape. . . or would mounting a tweeter to the back of the baffle and putting a big round-over on the front do the job?

 

[Scgorg]

I've looked at augerpro's. I think I'd need a 5", but it's available only in elliptical instead of circular. I'm a little confused about the positioning of a WG with respect to the midrange. If I understand correctly, the TM centre-to-centre distance is usually kept as small as possible, but I've also read that this distance should be 1.2 to 1.4 times the wavelength of the XO frequency. Trying to do both will push the XO frequency up quite high. Am I missing something? An elliptical WG seems to support keeping the distance small, but I've seen designs with circular WGs and quite large separations.

The 1.2-1.4 times wavelength at crossover is something Kimmosto, the creator of VituixCAD came up with. While doing this gives a relatively small vertical window for listening, it can give very smooth early reflections and sound power. The "worst case" scenario for sound power is to have the tweeter and midwoofer about 0.5 times the wavelength apart. The benefit of this arrangement is that the listener will be able to move more in terms of elevation while keeping good sound. I'd look at what the intended use case is and go from there.

As close spacing and low frequency xo as possible is something I would go for if you're having the loudspeaker quite close to you, such as on a desktop. The reason being that even relatively small movements can yield a substantial change in the listening angle relative to the loudspeaker. In addition, at close distances the impact of the overall sound power on the response is smaller.

For listening further away (say, more than 2 meters), sticking to something like 1-1.4 times wavelength at crossover makes more sense. It gives nice and smooth sound power, even for designs where the drivers are not coaxially located. Being further away also means that the smaller listening window is not such a big issue, as normal changes in seated elevation are gonna be a relatively small change in angle relative to the loudspeaker reference axis.

Edit: if you can get the drivers *really* close relative to wavelength, like 0.3WL or lower, then you pretty much get the best of both worlds. This is not practically achievable in the vast majority of cases.

 

[Momomo67890]

Keep in mind you can also cross lower with a waveguide tweeter.

 

[Old Hi-Fi Guy]

The 1.2-1.4 times wavelength at crossover is something Kimmosto, the creator of VituixCAD came up with. While doing this gives a relatively small vertical window for listening, it can give very smooth early reflections and sound power. The "worst case" scenario for sound power is to have the tweeter and midwoofer about 0.5 times the wavelength apart. The benefit of this arrangement is that the listener will be able to move more in terms of elevation while keeping good sound. I'd look at what the intended use case is and go from there.

As close spacing and low frequency xo as possible is something I would go for if you're having the loudspeaker quite close to you, such as on a desktop. The reason being that even relatively small movements can yield a substantial change in the listening angle relative to the loudspeaker. In addition, at close distances the impact of the overall sound power on the response is smaller.

For listening further away (say, more than 2 meters), sticking to something like 1-1.4 times wavelength at crossover makes more sense. It gives nice and smooth sound power, even for designs where the drivers are not coaxially located. Being further away also means that the smaller listening window is not such a big issue, as normal changes in seated elevation are gonna be a relatively small change in angle relative to the loudspeaker reference axis.

Edit: if you can get the drivers *really* close relative to wavelength, like 0.3WL or lower, then you pretty much get the best of both worlds. This is not practically achievable in the vast majority of cases.

I'm usually listening on my own in my recliner at around 3 metres from the speakers, so pretty consistent and relatively distant positioning. With the mid I'm considering, crossing at 2.4 kHz (say) would put me at 1.2 with a manageable space (about 3/16") between the flanges of the drivers. Sounds just right.

 

[Old Hi-Fi Guy]

Keep in mind you can also cross lower with a waveguide tweeter.

Why is that (only if there's a short answer)? Having some extra wriggle room may be useful, but doesn't a lower XO make the tweeter work harder? Hmm. . . I'd guess that the tweeter doesn't have to work quite as hard if the "acoustic impedance" (if that's the right term) of a flared pipe is less than for a squared-off pipe.

The question in the back of my mind is this: if wave guides are so good, why don't all tweeters come with them as a matter of course?

I still like the idea of getting out my big round-over bit. . .

. . . I suppose the only accurate way to determine the difference in acoustic centres between a mid and a tweet is by measuring their respective step response? I could then make my custom round-over the right depth.

 

[Momomo67890]

Check out the spl at 2.83v on augerpros graphs and compare to the datasheet. Waveguides till recently ain't easy to make and speaker company's are pretty basic advanced modeling is only something some have been using recently. Think of taking 180 degrees of sound and stuffing it into around 60. I use a bliesma 34b in a waveguide not augerpro that's flat from 1000hz to 20khz so it actually takes that 180 -> 60 and increases that 0-60 around 1200 by 8db but at same time at 20k it widens the beam from 40 to 60 and I lose dB so I've lost 4dB up high and gained 8 down low so on paper I have a less efficient tweeter but it has a much larger range it can be used in. I use it at 1600 and It can hit 120db on a sweep without blowing up but can also
Be used down to 1k. Another way to lower the xo is to use a steeper xo.

 

[Momomo67890]

SB17CAC35 is a excellent mid for its price. check out the review for it on erins audio corner its got -60db thd. For mid range drivers le is the most important parameter.

 

[Old Hi-Fi Guy]

SB17CAC35 is a excellent mid for its price. check out the review for it on erins audio corner its got -60db thd. For mid range drivers le is the most important parameter.

Thank you! My proposed speaker is a 3-way, so the SB17 would be overkill. Erin suggests that the SB15 would be a better choice for 3-ways. There is also an SB12 which is a 4", but its Le is 0.34 mH versus 0.14 mH for the 5", so perhaps the 5" would be the better choice?

I'm somewhat deterred from using the SB ceramic cone drivers by their harsh break-up zones, but getting a lower XO frequency would certainly help. I was intending to use a Seas 4" which has a smooth, extended FR, but its Le is 0.31 mH. There appears to be a lot of trade-offs.

Your Bliesma has a spectacular FR!

 

[Old Hi-Fi Guy]

Since I don't like to leave these things hanging, and in case anyone is curious:

I decided to go for the Seas MCA12RC mid and the SB Acoustics SB26STAC tweet to accompany the Seas W26FX001 woofers. I've worked up a design for a mini Stradivari cabinet, which looks quite promising. It's "mini" in terms of its reduced height when compared to the Sonus Faber original due to my use of a single 87 dB woofer. The width, depth, and general profile will be similar. The hardest part so far has been explaining to my wife why we need another pair of speakers.

I have rough-cut most of the parts from two sheets of absurdly expensive German birch plywood (Baltic birch was even more expensive). I've had a sheet of fiddle-back maple veneer languishing in my workshop for several years waiting for a project - I just have to try not to screw up the finishing like I usually do.

 

[Dave Bullet]

My mains 3 ways are Seas 27TDFC, L15 and L26. The L26 is in a ~70L box sealed. F3 around 42Hz.

I play these in a small/medium sized 5m x 4m living room. They do not have the bass extension I like. I always play them with my old trusty Shiva 12" 6 cubic foot sealed sub - it integrates well.

The L26 can take a beating with large Xmech, but only linear to 7mm. If you want bass extension without a sub, especially in a larger listening room - go ported with ~ 90L (3 cubic foot) enclosure.

 

[Old Hi-Fi Guy]

My mains 3 ways are Seas 27TDFC, L15 and L26. The L26 is in a ~70L box sealed. F3 around 42Hz.

I play these in a small/medium sized 5m x 4m living room. They do not have the bass extension I like. I always play them with my old trusty Shiva 12" 6 cubic foot sealed sub - it integrates well.

The L26 can take a beating with large Xmech, but only linear to 7mm. If you want bass extension without a sub, especially in a larger listening room - go ported with ~ 90L (3 cubic foot) enclosure.

I do like good bass, but these speakers will end up either in my listening room, in which case they won't see anything much lower than a string bass or piano, or in my HT system. The latter is unlikely unless I either build passive crossover for them eventually or make my HT system active. The mains in my HT system have 11" Eton woofers in ported boxes that have an F3 approaching 27 Hz, but that's because I built them before I had a sub. From memory, I think they're in around 65 litres. Now that I have a sub, I could go smaller, which would please my wife. They're approaching twenty years old and the boxes need refinishing. If that turns out well, I'll keep them; if not, there could be another speaker project in my future.

According to Unibox, my Mini Stradivari should have an F3 of 42 Hz - sealed in 45 litres. A bit lower would have been nice, but this is a good compromise and seems to be working out well so far. The first baffle is in my shop-built vacuum press at the moment - time to check on how it's doing. . .

 

[Old Hi-Fi Guy]

Well, four months have gone by, punctuated by a few set-backs, a fair amount of frustration, and copious amounts of swearing. The finished speakers are now in my listening room waiting for me to get around to doing the DSP stuff. I don't know when this will be, because I have a lot of other work that needs my attention and this project has already taken far longer than I expected. Anyway, here's a photograph:

 

[Dave Bullet]

Beautiful looking speaker!

Did you put the midrange in its own sealed sub-enclosure?

Did you end up with a passive or active crossover?

 

[Old Hi-Fi Guy]

The box looks generally like this inside, so yes the mid has its own enclosure. It's just a box with drivers in it at the moment. The next step is to test and configure DSP crossovers.

 

[kemmler3D]

if wave guides are so good, why don't all tweeters come with them as a matter of course?

Waveguides are somewhat application-specific, so selling tweeters with waveguides is a little bit like selling wheels with tires already on. They may work for a large percentage of users, but never all.

Designing a waveguide is also another design and manufacturing step that the speaker manufacturer usually handles for their own specific application. So selling tweeters with WGs would probably only help a certain segment of the DIY audience, not a huge one by all accounts.

 

[Dave Bullet]

What are you planning for measurement? Getting a microphone would be my first step

 

[Old Hi-Fi Guy]

I have previously used REW with a miniDSP and UMIK microphone. I'll do the same this time, but first I have to find the time to re-learn it all. Also, as we roll towards summer, the weather should allow me to do some measurements outside, so I'm not in any hurry even though I can't wait to hear what they sound like. No doubt I'll end up doing most of it indoors and then fill in the bass outside.

 

[terryforsythe]

I have previously used REW with a miniDSP and UMIK microphone. I'll do the same this time, but first I have to find the time to re-learn it all. Also, as we roll towards summer, the weather should allow me to do some measurements outside, so I'm not in any hurry even though I can't wait to hear what they sound like. No doubt I'll end up doing most of it indoors and then fill in the bass outside.

I don't see a reason to wait. I used measurements made outdoors for initial modelling to set initial crossover and EQ parameters. But, once I had the speakers put back together and started tuning the speakers for the room, the initial crossover and EQ settings went out the window. I optimized the tuning for the room and my listening preferences.