18W, 12M and R3004 active 3-Way

[DannerD3H]

Hello people!
Welcome to my build log (long term..) of my new active DIY speakers.

6 years ago I build a pair of stereo speakers and a subwoofer. I bought the drivers I thought looked the coolest and build some simple boxes that I thought looked good without thinking much about it…. They ended up like this (they now have the R3004 tweeter):

They play really well! But, now when I am a bit older I know that l do not do the drivers justice in this cabinet, at all! I think I can do better! And I have gotten a little tired of the white cabinets xD

So.. I am starting over and will think it through this time. At least as much as I can. I already did a lot of reading and simulating with WinISD. However, I am still a rookie when it comes to speaker building… Therefore, with this thread I hope to get a sanity check of my design and get some tips along the way, when I start building. And of course share my own ideas and findings with the rest of you!
I always really wanted to have a three-way speaker. Partially from the look but also to have more freedom in the bass level compared to the mid. When spending this much time and money on speakers I want them to also look good, therefore I have chosen the matching (but ridiculously expensive) 12M to match my woofer in looks and quality.
I always wanted to try active crossovers because I think it makes most sense for me. All my audio sources are digital anyway (PC, Phone, TV, Xbox). I also like the flexibility it gives and I think it makes more sense than passive (I do not wish to start a passive VS. Active war ) And now that Hypex have launched their Fusion Amp series there is no doubt for me! I can then connect my TV to the master amp with an optical cable and then switch between my HDMI sources on the TV. The volume can then be controlled with the Hypex remote. I can also connect my Yamaha digital Bluetooth receiver to the coaxial input on the master. Finally the master and slave is connected with a coax. This will give a 100% digital system right until the amp for each driver. I think this will give a pretty sleek system with only one D/A conversion. Today I use a rather cheap 125W DENON 7.1 AVR.
I have looked for inspiration on the net. I have been looking a lot at Troels Gravesens designs as he have used the same drivers in a few of his builds. The Ekta is a good example along with the Elam flex. However, I have looked at everything I could find where people use the 18W. I have also found a lot of inspiration in Sonus Fabers speakers with respect to looks.

My current new speaker design:

Some facts:
3-way 2.0 stereo setup driven by 2 Hypex FA123 (Fusion amp with two 125W channels and one 100W tweeter channel) one in each speaker.
Scan-Speak 18W-8531G00 woofer crossed at 700 Hz
Scan-Speak 12M-4631G00 midrange crossed at 700 and 2500 Hz
Scan-Speak R3004/662000 Tweeter crossed at 2500 Hz
28 L main cabinet (actual volume with bracing subtracted).
1,5 L mid cabinet (based on the Ekta).
22 mm MDF for internal parts, bottom and rear. 44mm front baffle and top made of two 22 mm layers
24 mm sides made of four 6mm MDF layers bend to shape.
Ø66 port with 15 mm flare.

I have designed the speakers so that the baffle, rear, top, foot and back plate are all mounted with bolts and threaded inserts. The wooden outer side panels I also want to be removable but haven’t decided how yet. I am actually considering some really strong Velcro I have lying around. The reason for this is that I would like to be able to change the colors over time.. It may be stupid, but I really think it is possible and if not then I can always glue it.. it is harder to go the other way
At this point I plan to either cover all black parts in faux leather or use black stain on veneer. The only issue with this is that I would then have to make the front baffle of a piece of actual wood to get a proper looking result – which scares me.
The wooden parts will be birch with a slightly darkening oil and then polished with wax.

This is just a small teaser. I can promise that a million questions will follow. I plan on starting the cabinet build this Christmas.. so still some time to get everything settled.

 

[eriksquires]

Heh, thanks for the tip on the Hypex modules!

Those 3-channel amps are ridiculously reasonably priced, and make going with passive or a hybrid crossover (passive for tweet and mid) unnecessary.

Best,

E

 

[DannerD3H]

Haha yes they are very well priced! and they look really interesting!

 

[DannerD3H]

Cabinet size and simulations

I have chosen the bass reflex type because I like bas. I like tight bass the most, but a sealed cabinet with these drivers are just not deep enough for me.. I also considered to use two 6 in passive radiators in each speakers, but didn’t dare and from WinISD it didn’t seem it was worth the extra cost.. I never heard passive radiators in real life anyway..
This is how the speakers look on the inside:

The cabinets have an actual internal volume of 28L and the ports are Ø66(internal) x 230 mm. The volume is a compromise of the design/size and performance. And I am basing it on WinISD simulations. This is what the transfer function looks like:

My goal was to get an f3 under 40Hz. The simulation includes a high-pass filter to protect the woofer from over excursion. It is a second order filter at 28 Hz. This is based on a target maximum volume of 103 dB at 20 Watts.
I have been told this slight bump is desirable but I do not understand why. Can anyone explain that?

Troels Gravesen states that real life ports are usually only 60-70% the length of calculated ports. Is this something anyone here can relate to?


Here are some more graphs (Group delay, Cone excursion@ 20W and port airspeed @ 20W):

 

[Rokytheman]

Interesting!

If I were you, I would build the cabinet of baltic birch plywood. It is nicer to work with - And if you first start on reading up on cabinet materials - All evidence points in a direction where MDF is worser than baltic birch plywood.

Also I would think about if one 18W is enough. Yes it can play quite som bass - And the quality can be outstanding!!! But not very high SPL levels. Especially if you listen to music with a lot of bass.

Also - In regards to tweeter placement. Simulations are fine, but I will advocate to try some placement on a baffle before your final decision. Cardboard is fine for testing.

For bass you could also consider SB Satori 9 1/2" (have not heard this one, but for sure a very very nice driver !!!) and midrange the Satori 6".

 

[Rokytheman]

I also think you should consider making some kind of aperiodic enclosure for the midrange. From my experience it sounds better than a closed cabinet. Always.

 

[DannerD3H]

Hello Roky!

I really seem to find contradicting information regarding the MDF VS. Ply cabinets.. Using Plywood would allow me to use black stain instead of using leather without having to add veneer. So i would diffinitively prefer using plywood!.
Which is more bendy? MDF or Plywood of the same size?
And also, i see that you too are a dane. Is what we call "birke krydsfiner" always baltic birch or can it be something else too?
Odense Lavpris Tømmerhandel A/S

Those SB drivers sure also look good! However, i already have the 18W and the tweeter, and i would really like to keep them

About a second 18W:
If i use two of these wouldn't i effectively need double the cabinet volume?
Are you sure it cannot play loud? From the simulations of SPL in WinISD it looks perfectly fine(@ 20W)...

Jeesee... Everytime i think i have things sligltly under control something new comes up.. Am i understanding it correctly that it is basically just a vented cabinet for the mid?

 

[Rokytheman]

DannerD3H: I think about the same. But I will bet that plywood can be bend more. But If you want to bend it without cutting it first, you will need to use 4 mm sheets. But 15 mm plywood cannot just be bend.

Yes Im also from Denmark - It is called "birkekrydsfiner". There are several qualities - Or normally just two. The good one has a very nice front - With a nice finish. You are not in doubt when you see it. Silvan has it in both qualities. I think it is also called "BB" quality. The good quality has the same kind of layers, where the cheap version has softer layer in the middle. Keep in mind that plywood need more care when cutting - It can chip in the corners.

I have thought about using a cheaper quality and combining it with the quality version on the outside. I think that constrained layer damping is the ultimate solution.

For the bass you want the cabinet to be as stiff as possible. For the midrange it is not the same. Here you want to dampen the resonances. Ie - a very stif cabinet will have resonances higher up in frequency. Sadly, this complicates the building process haha.

I think MDF is typically chosen over plywood, because it is easy to work with and it is cheap. Not because it is better - Also it is a mess to work with, compared to plywood.

Several years ago I built Troels Gravesens SP95. I cant remember how loud I played, I just remember that I could get it to button out which is not a nice sound or experience- And it was not insane SPL levels. So my point is that if you like to play loud and or you listen to music with a lot of bass, you probably should consider a bigger driver.

Yes you can see it as a controlled vented cabined - but not with a reflex port. You can google it. Also named as "flow-resistor". Scan Speak makes one that are ready to use, but they are expensive. Dynaudio invented it

Keep in mind there are many different opinions about how to make a good speaker. Mine is just one :-D

 

[scudinc]

The 12M has a very simple motor for such an expensive driver. I'd go for the 12MU instead.

Have you simulated the baffle diffraction? Those offset drivers make me go cross-eyed.

 

[DannerD3H]

Okay, then i will have to use plywood. I can unly find MDF in 6 mm and up. And i will prefer not to cut it to have maximum strength. No i will make the sides of more smaller layers.

Yeah okay so it is the quality that defines if it is good or not. not the location. Will have a look in Silvan then!

hmmm i think i am already in deep water with this cabinet design.. I have no idea how to determine the proper size and so on for the flow resistor.. I will definitively have to read mor about that before i can say if i can do it or not.
Can you descripe how it sounds different?

I never had them bottom out in the current cabinets. And as far as i know i have no protection against it. The new Speakers will be controlled via DSP and thus i will put in a highpass filter to protect them.


I know, speakers and sound are very subjective

 

[DannerD3H]

Scudinc,

That is actually a good point.. The illuminators cost the same.. That will probably also match the tweeter better!

No i haven't. I don't think WinISD can do that. Can it?
Do you think this position is gonna make things worse than having them centered?

 

[cowfoot]

@DannerD3H,

Try having a look at Boxsim from Visaton, it's a piece of software like WinISD, but it's able to simulate difraction and baffle step. You can find it at boxsim.de.

There is also VituixCAD, it's able to simulate alot of things aswell. It can be found here, Software.

 

[eriksquires]

One suggestion of course, since you are going fully active, is to use a sealed, undersized cabinet with EQ to flatten the bass. You'll get more efficiency and lower extension, not to mention reduced cabinet size.

I'm afraid you may need to rely on others for more info, that's all I got.

Best,

E

 

[jReave]

@DannerD3H,

Try having a look at Boxsim

Does Boxsim model anything besides strictly rectangular boxes?

If not, I think the best free program for your use is the Baffle Diffraction Simulator as it will do oddly shaped 4-sided boxes and simulate rounded/chamfered edges at the same time. It needs Excel though and it's not exactly the most intuitive program to start off with. And it won't be able to model your design perfectly but it will do a reasonable imitation of what's going on at the top with your tweeter and mid.

Example below with a 41" tall cabinet with 1.5" chamfered edges with 7" top width and 16" bottom width (not your cabinet shape I know but we're only trying to simulate the top shape around the tweeter and mid). From the bottom tweeter is 36.5" and the mid is 32". And with that shape you can see that there is no need to offset either the tweeter or the mid from center - baffle diffraction ripple is well controlled above 3000Hz for the tweeter and above about 1700Hz for the mid. These types of sims should be good to get you started but for this type of baffle, I second the cardboard method. But you need to be able to take measurements. Pretty much a given actually for a project like this.

As mentioned previously, there are many personal preferences. To better absorb the backwave for the mid, I would extend the chamber deeper, perhaps right to the back, and then use increasing densities of insulation as you go further from the driver.

To expand on eric's idea above - since you are going active you have the ability to boost your bottom end in which case you could use 2 of those mid/woofers sealed in as little as 30L. They won't go as loud as low as vented but you will get about 98dB at 40Hz before xmax is exceeded. Better transients with sealed too. In which case I would also design a deeper cabinet again and again use a lot of high density insulation to absorb more of the rear energy.

Or just another random thought - you could use the 18W as your mid and then go with 1 or 2 larger woofers (also sealed?). That would be a pretty different speaker though.

 

[DannerD3H]

#cowfoot - Thanks for the software tips! They look alittle more frightening than WinISD so i will have to spend some time in them before i get something useful i think..

#jReave - I will also look at that! seems more straight forward for diffraction at least!. However, i have no idea what to look for on those graphs and what is good enough. The graphs you show is that a somewhat realistic target? I guess the ideal is flat as usual?


The problem for my design is that I really would like to not make it even bigger. My living room is not very big.. I do not wish to have a bigger woofer in the cabinets. However, i may rebuild my 30W sub (the one i have build right now is really nonideal..) at some point.
Letting the mid-cabinet go all the way to the back will cut the top room off and effectively reduce my woofer volume from 28l to 23 l. Unless of course i doo something fancy..



#eriksquires and #jReave - Your idea of a smaller cabinet going sealed sounds very interesting to me!
Can you elaborate alittle on this for me, or point me to some interesting reading regarding this kind of bass boost?
Will it be possible to make a sealed cabinet of < 28L with 1 or 2 18W and stil reach a f3 around 40hz using some kind of wizzadry in the fusion amps filtering? I am just used to the fact that nothing comes without a cost in speaker buildings.. But this sounds like a pure Win?

 

[eriksquires]

Also keep in mind room gain. A -3 dB at 35-40 Hz speaker may be essentially flat to 20. With EQ, yumm yumm.

 

[jReave]

Right then. I've just had a quick look at those Hypex amps. It appears there are no options for bass boost as there would be using dsp so that idea may be out unless you choose to do it by some other means.

Also there appears to be simply three preset types of xo filters to choose from with the Hypexs (I'm assuming that they are xo filters). For the quality of drivers you are selecting, I would certainly want to know a whole lot more about those filters beforehand. What are the xo frequencies? What order? What type? Is there baffles step compensation included? What about getting proper phase alignment between drivers? I'm not really that knowledgeable about active nor have I worked with these specific drivers before (the Revelator15W - yes), but this is not the route I would go unless I was also willing to either passively or perhaps digitally fix any problems post (or pre?) Hypex filtering.

And for that you need to measure. Just wondering then if you plan on doing that as some people think that going digital allows them to bypass this most important step?

Re diffraction - play with a program to start to understand what's going on. The Edge may be the simplest and fastest to get up and running on, although it doesn't allow any kind of edge treatment simulation. If you have Excel, try Bagby's BDBS for that.

Typically there are 3 aspects to diffraction - a standard 6dB loss on the bottom end, a small peak somewhere in the middle and then ripple in the higher frequencies. Edge treatment tends to smooth the ripples, while baffle width determines the frequencies where the peaking and the subsequent 6dB loss happen. What's important is for the driver's FR and the driver's baffle diffraction response to sum as smoothly as possible within each driver's frequency range that's being used.

So for the tweeter (blue), my sim above shows minimum ripples above 3kHz which is excellent. Between 1kHz and 3kHz, there's a 1dB dip which should work well with your tweeter selection since it has a small peak in about that same frequency range. Below that isn't too important as it's below the xo frequency and the xo filter will usually take care of that. Note this is just a simulation. Your baffle/chamfer style is a good one but very difficult to sim correctly. For these drivers, accurate cardboard mockups and measuring would be the best way to dial in the flattest overall response. Then build the cabinet.

 

[eriksquires]

Ahem:

From here:

Hypex Electronics B.V.

I quoteth thus:

5 biquads per amplifier

 

[Doppler9000]

One suggestion of course, since you are going fully active, is to use a sealed, undersized cabinet with EQ to flatten the bass. You'll get more efficiency and lower extension, not to mention reduced cabinet size.

I'm afraid you may need to rely on others for more info, that's all I got.

Best,

E

I would avoid the undersized-box approach.

Professor Linkwitz recommends a Qtc in the critically-damped range.

There is no free lunch, even with DSP.

 

[jReave]

Hey eric, actually it quotes 15 biquads per amplifier, but that's Greek to me.

What's a biquad and what can it do?