Seas 3-way wide baffle

[norcad]

Hello, I'm new to this forum and after reading about Dave Bullet's Seas all metal 3-way, I thought it could be fun to make a thread about my own DIY 3-way Seas.

First of all, I'm Norwegian, and to write in English is not easy!
So please tell me if there is something I write you dont understand

It all startet 18 months ago, when I decided to build my own speakers for my home theather. I have some experience from building car stereo, but this was my first DIY "hifi" speaker.

I was getting help from a norwegian forum and learned a lot in the beginning:
- You need measuring equipment!
(I bought a ECM-8000 mic, and M-Audio soundcard.)
- Without experience you should not use metal drivers!
(Paperdrivers it will be.)
- Passive XO is very difficult, so you should not start with a 3-way!
(But I want bass!?)
- You must remember baffle step when your looking for drivers!
(Baffle what?)

When I understood what baffle step was, an idea was born.
I wanted to build a 3-way with an active xo for bass/mid and passive for mid/tweet. The baffle should be wide enough to push the BS down 2-300hz.
Because I'm from Norway it couldnt be any other drivers than Seas of course! :T
I choose the 27TFFC, CA18RNX and CA26RFX

After some discussion with my wife, I ended up with this drawings:





I used the "Baffle Diffraction Simulator" and this is the result:



Many people told me that a small baffle is much better, mostly because of diffraction. Well this is a normal shaped box - 30cm wide:



Driver 1 (red)=mid
Driver 2 (blue)=tweeter
Driver 3 (yellow)=bass

More to come.........

[Jed]

Keep it coming Norcad! :T

[Landroval]

Looks good! I don't know if you know this, but might be of interest:

http://www.troelsgravesen.dk/PMS.htm

[Winter]

The wide baffle with sloped back ends, as you are aware, has several benefits over a narrow baffle, though not very often mentioned. It's basically a portable infinite baffle design. Low edge diffraction , and much more even power response as the baffle step has been pushed down into the region of room wall gain. Several listeners have mentioned it sounds similar to a dipole.

Commercially, Sonus Faber builds the Stradivari and Elipsa wide baffle speakers. And don't forget the TAD Model 1.

http://www.troelsgravesen.dk/Acapella_WB.htm

http://www.troelsgravesen.dk/PMS.htm

http://www.sumikoaudio.net/sonus/prod_elipsa.htm

http://www.stereophile.com/floorloudspeakers/1207sonus/

http://www.sumikoaudio.net/sonus/prod_stradivari.htm

http://www.stereophile.com/floorloudspeakers/105sonus/

July 2004 Issue of The Absolute Sound 'Sonus Faber Stradivari "Homage" Loudspeaker' (no link)

http://tadhomeaudio.com/model-1.html

[norcad]

With those dimensions (as you see on the drawings) the bass-enclosure will be 64L, and the mid-enclosure will be 5,4L.
When I ran some simulations in Bassbox, this should give me -3dB@40hz, and a qtc of 0.707 for the mid.




Not all that low bass I was hoping for, but hopefully good enough for music listening without using the sub.
The sub is btw a DIY Peerless XLS12 in 45L closed box, driven by Hypex DS4.0 with eq.

I also bought a pair of Seas L26, who has a lower fs, but it needed a bigger enclosure, and that was not my wife very happy about!

In the meantime I was trying to learn Speaker Workshop, testing Arta Software and LspCad, made a "Wallins jig" (that didnt work!), and made a couple of testboxes to measure TS-data. In the middle of all this, something called acoustics was foolin around with me also 8O

For a beginner this took hours, days and weeks of my spare time.
After a couple of months I finally bought 22mm MDF and was ready to start building the enclosures.

[TacoD]

I like your enthusiasm, I think lspCAD is a good start way easier to understand than Speaker Workshop.

-3 dB @ 40Hz is enough output for medium rooms, do not forget the extra output a room adds to your loudspeakers. Most loudspeakers (also the big one's) do not reach that far down.

[owdi]

A few recommendations....

For midrange, use a driver designed for midrange use, such as the MCA15RCY. Compared to the CA18RLY, this driver is more efficient, more sensitive, has better off axis response at your crossover frequency, and the effective acoustic depth is shallower. These all contribute to simplifying crossover design.

You can save a lot of work by using these MDF L-corner custom curves, although I don't know if rockler.com will do international orders. Radius is about 7.6cm.

http://www.rockler.com/product.cfm?page=10569&filter=wood%20and%20veneer& cookietest=1

This is what they look like unfinished.
Image not available

Dan

[norcad]

Thanks for your reply, and please continue with it.
This speakers are already (almost) finished, but it wont be my last so every tip and tricks can be useful

"owdi": The midrange driver is the CA18RNX and not RLY. The reason for this is that I wanted to use the same drivers in center and surrounds, without the CA26 of course.
But please continue to follow my work, maybe you will be surprised

"TacoD": I tried LspCad, but I'm still using Speaker Workshop. Its free!

[Dave Bullet]

Hi Norcad / all,

Great to see the progress.

I've been tossing up between a narrow baffle for my midrange, vs a wider baffle. My concerns are around imaging. Does it help to have a narrow baffle to improve imaging or is this just a fallacy?

Is imaging mainly a function of really good driver integration (ie. phase tracking through the crossover?)

Going wider baffle will help my design - making the enclosure easier to construct and raising the L15 high pass slope for L26 integration.

The only other concern I'd have is whether there will be increased backwave reflection on the drivers possibly colouring the midrange with a shallower cabinet (23cm overall)?

David.

[Bastek]

Imaging is not so good with wide baffles, but the soundstage is so big and believable, it goes closer to the live event in my opinion.
My slim tower baffle is 9"x50" - great "studio monitor' sound.
My wide tower baffle is 18"x60" - great "live music" presentation.

[norcad]

Hi Norcad / all,

Great to see the progress.

I've been tossing up between a narrow baffle for my midrange, vs a wider baffle. My concerns are around imaging. Does it help to have a narrow baffle to improve imaging or is this just a fallacy?

Is imaging mainly a function of really good driver integration (ie. phase tracking through the crossover?)

Going wider baffle will help my design - making the enclosure easier to construct and raising the L15 high pass slope for L26 integration.

The only other concern I'd have is whether there will be increased backwave reflection on the drivers possibly colouring the midrange with a shallower cabinet (23cm overall)?

David.

I'll think you should go to Troels site, and read about his PoorManStrad (www.troelsgravesen.dk). I can just say that I agree with him, and that IMO a wide baffle has many other benefits that for me maybe is even more important. My speakers has great imaging horisontal, but lacks a bit in depth and hight, but I'm not sure my cd player and amp would do it any better with other speakers.

The backwave reflections can be dealed with. By stuffing, angled walls and so on. My paper midrange are even more sensitiv to that than your metal cone. Standing waves in the mid enclosure can be a problem so take your time and do a little math, and try not to have two dimensions in the enclosure that is the same.

[norcad]

Finally it was time for some box building!

The first challenge; how can I make those big radius's in MDF?

After a lot of thinking, drawing and a little math, I had decided that the radius on each side of the baffle, should be made out of 5 pieces, and have an inside radius of 10cm. (I don't know what to call those "pieces" in English?)
Every "thing" should be 3,9 cm wide, with an angle of 9 degree.

Well, look at the pictures, they say more than 1000 words of bad English! :lol:

Here you can see 5 of those ????, and the PVC pipe who will be used as support:






A little test, to see if my math was correct:






And here you can see that 20 of those, makes a nice MDF pipe






And here they are glued together, with the baffles and sides as well:






The first impression of how it look IRL:



The next challenge coming up; bracing!

[norcad]

So far, so good.
I was afraid that making the bracing should be nearly impossible, there was many angles that had to match very good. Fortunately it went easier than expected. Beginners luck?

Here are the raw bracings starting to match up:




And the mid-enclosure, with its tilted backplate:




The bracing has got some hole, and the bass driver are getting good support:




Bass support close up:




And finally, all bracing and midenclosure glued to the frontbaffle:

[TacoD]

Nicely done, you have some serious woodworking skills .

[norcad]

Thank you, TacoD

So lets get those cabs finished!

Those big radius "corners":




The side panels:




How it looks so far:




And then we take a jump in time (and work).
Top and bottom are cut and glued, and all three drivers are in place.




And here we can see Puss (without boots!) testing the new "house".

[Jonasz]

Looks awesome norcad! :T

Var i Norge bor du? Provlyssning? :B

[norcad]

Thank you Jonasz
Jeg bor ikke så langt fra Lillehammer (Vinter-OL 1994).

OK, the cabs starts to look like something, and here are even more pictures! (I hope you enjoy pics more than just boring text!)

The back plate is still missing, so first I glued on some small MDF pieces around the inside edge of the hole cabinet, mostly because I need to get inside the box to test some stuffing, different lengths on the vent, and of course a lot of crossover tweaks.




And here is the backplate done:




The 3" vent (hopefully big enough):




And then a HUGE jump in time and work, the other cabinet nearly finish:




Here you can see how the long vent can be from the inside, if I want to test the L26:

[Landroval]

Looks very good again! Are the double terminals for bi-wiring, or what is the story behind them?

[norcad]

Looks very good again! Are the double terminals for bi-wiring, or what is the story behind them?

Thank you! (again )
They are for bi-amping. I'm going to use an active xo for bass/mid, and passive for mid/tweet.

[TacoD]

How thick is the back, 18mm? Are you not worried for vibrations direct behind the woofer (only 1 brace, it's in the middle of the panel, but it's a wide speaker ).

[complet]

Looks great!

[norcad]

How thick is the back, 18mm? Are you not worried for vibrations direct behind the woofer (only 1 brace, it's in the middle of the panel, but it's a wide speaker ).

Its 22mm MDF like the rest of the cab. And no I wasnt worried about it, before I got the same qt at the norwegian forum.

The biggest part of the backplate without support is 22x32cm, not that big actually. It also has a layer of 4mm butyl and 5cm stuffing. I havent noticed any problem with that. The only place I can feel vibration on the cabs is on the topplate, on each side of the midenclosure. I hadnt expect that either, but I suppose this is where the pressure inside the box is highest.

[norcad]

So far, so good. Now it was time for something I never had done before.
Measurements and crossover.
Be aware of the fact that I'm a beginner to this, so it might be some errors in the measurements, at least those from the beginning of this project.

First, a nearfield measurement of all drivers and port.
Red= 27TDFC
Blue= CA18RNX
Black= CA26RFX
Green= Port




Here you can see the simulated bass responses from Speaker Workshop vs measured response. The measurements are nearfield with driver and port summed.




And some measurements from different distances
Green = summed nearfield from driver and port
Black = at 10cm, on axis bass driver
Red = at 1m, on axis bass driver
Blue = at 2m, on axis bass driver
Yellow = at sweet spot, in listening height.

[norcad]

Its time for some updates.
The bass respons was nearly spot on, something the impedance measurements stated as well. The tuning was a dB or two on the low side, but that's OK.
Now it was time for the passive xo, and I started with 2nd order LR @ 2,5khz. I soon realized that this didn't work, the phase tracking was poor and I couldn't get any reverse null, because of the flat baffle.
When I put an resistor in series with the cap on the midrange, so it looked like a first order with zobel, and changed the values on the tweeters xo, so I got 3.order roll off, things started to look much better.

Looks better, yes. But sounded, well not so good

After a lot of measuring, tweaking, trying different xo points and slopes, I tried to measure the speakers with the xo's in Arta Software.

And it looked like this:

O naxis with xo, tweeter, midrange, summed and reverse null, all with 1/3octave smoothing:




Measured at 0-15-30-45 degree.




And reverse null without smoothing




Can YOU see any problem? Do you think this should sound great?

[Dave Bullet]

Hi Norcad,

The relative dip between 700Hz and 2KHz might make the speaker sound a bit recessed (female voices). Depends on how you like your mids.

Looking at your BDS sims and guessing baffle step - seems the midrange is the root cause to the 700Hz - 1KHz drop. Without changing the baffle (doesn't look like an option since the bracing relies on the mid being centred), you could try reducing the mid padding resistor and shaping the HP knee (higher Q) of the midpass so that it raises the 700 - 1KHz area. If that makes the mid too peaky around the 1 - 1.5KHz area - try using a capacitor / resistor parallel shelving pair to lower rather than just a series resistor in front of the first cap (look at my 3 ways for an example).

Of course - I have yet to build a pair of speakers I would call reference... so take my advice with a grain of salt.

Cheers,
David.

[norcad]

The problem was that in the beginning I used Speaker Workshop, and with a short gating it uses a low resolution in the measurements. The dip was indeed visible, and I used days and weeks trying to find out why the dip was there.
After a couple of weeks, I finally did two things that solved it. First of all I tried Arta Software, who showed me that it isnt just a dip around 1-1.5khz, but a peak around 6-700hz! The second was that I bought an eq, and started to adjust the sound.

With a notch filter at 700hz, it started to sound real good

[norcad]

I started with electric 2.order, and tried different xo-points, everything from 1800 to 2700hz. Without any experience, I made som goals out of what I had read, one of them was phase tracking and reverse null, an another was electric phase.

Here is some of my first attempt:









They all had one thing in common, I didnt like the sound!
As soon as I turned the volume up, the speakers start "screaming", the sound was "hard" and my ears was tired after 10-15 minutes.

[Dave Bullet]

The 3-7Khz peak might have something to do with it.

What is the CA18's harmonic distortion like above 1800Hz? Would you say any one of those 3 crossover points you tried was better? In other words, if the 1800 was better than the 2700 - yet the overall FR curve is the same - it might hint at CA18 distortion at the upper end of the midrange causing problems. Canyou please post individual measured driver responses with crossover in place?

I don't think padding is your solution - sure it will "dull" the harshness, but it will dull the overall sound and recess vocals and the like.

I think a flat response can be quite acceptable. I have a pair of KEF C30's that measure reasonably flat and are not as harsh as another different pair which also measure flat. I think it comes down harmonic distortion. Forget about power response... I'm talking harsh no matter where you are in the room, whereas I can listen to the KEFs for hours no matter where I am in the room. A harsh speaker is a harsh speaker in my book, regardless of power response.

David.

[norcad]

I dont know how to explain the sound I didnt like, it was always there with all the xo's, more or less maybe? I had a lot of ideas, and some of the best DIY'ers in Norway tried to help me, and a lot of things was suggested.

Here is what Zaph writes about the CA18RNX:
This Seas standard line paper cone driver has one of the flattest and most extended response curves in the group. T/S parameters are optimal for medium size vented enclosures. Excellent bass performance. Slightly higher 2nd and 3rd order HD in the midrange, but low tall order harmonics. Overall a well balanced and sensitive paper driver

[HareBrained]

I don't have any idea as to why these aren't sounding good but a couple of simple things to try would be:
1. Try listening far off axis. This will decrease the impact of the 10k-20k rise. If the harshness decreases look to adding a notch.
2. Move the speakers closer and further from the wall. The in-room response may be coloring the sound and the xo might have to account for it.
3. Mount the CA18 on a separate baffle. It's possible the midrange enclosure is too small or the CA18 output is back-driving the tweeter.

Again, these are just things to try and I don't know how they'll affect your sound.

[Winter]

Below is a chart from Martin Collom's "High Performance Loudspeakers" book that may be of help:
_________________________________________________
COLORATION

Boomy, 50 - 80 HZ
Chesty, plummy, 100 - 150 Hz
Boxy, hollow 150 - 300 Hz

Tube like, tunnelly 400 - 600 Hz
Cup like, honky, 700 - 1.2 kHz
Nasal, hard, 1.8 - 2.5 kHz

Presence, upper hardness, wiry 2.5 - 5.0 kHz
Sharp, metallic, sibilant 5.0 - 8.0 kHz
Fizzy, gritty, 10 -15 kHz
_________________________________________________

[Dave Bullet]

Norcad,

Could it be a backwave reflection through the cone? I don't know how that would affect the sound (given the FR curve is flattish).

Can you listen to each driver individually with that section of the XO in place? It might help pinpoint whether it is one particular driver contributing most of the disliked sound.

I like the idea of the other poster suggesting taking the CA18 out of the current enclosure. Playing that driver alone (in a temporary / deeper and well stuffed enclosure) might prove / disprove a cabinet impact on the sound.

David.

[norcad]

I tested out many different things, like different stuffing material, tried another amp, another cd player, and moved it to an another room and so on.
And one day, I just started over. The mids enclosures where almost filled with dampening, SW was calibrated, new measurements were made, and new xo's was constructed.
This time I tried to get 3rd order slopes, but still with a reverse null, and a notch filter around 700hz. The xo point was 2500hz.

And here's a measurement in Arta showing the result.




for one reason or another this sounded much better, but on some recordings there was some sibilants, so I also tried a notch around 4khz on the tweeter.
The result of this notchfilter are seen here:




Now I finally could sit back and listen to the music, even at higher volume!

[Winter]

Norcad,

Some things you may consider,

1) Is your ECM-8000 microphone calibrated? The response curves vary for the ECM-8000. My ECM-8000 mic was sufficiently accurate without the calibration (after receiving the calibration curve), but I have seen several other calibration files for ECM-8000's that revealed that their calibration was essential.

2) That 4 dB peak in the 11 kHz to 17 kHz range is a little high for most people. Everyone varies in their upper frequency tuning preferences. Depending on the tweeter, I aim for flat to -2 dB in the 10 kHz to 15 kHz region, measured 10 to 15 degrees off-axis.

On another matter, have you considered contouring the top edge of the baffle the same as the sides? You have spend so much effort with good results on the front baffle shape to minimize baffle diffraction and baffle step. The front baffle top edge is fairly close to the tweeter, and not too far from the midrange.

[norcad]

Norcad,

Some things you may consider,

1) Is your ECM-8000 microphone calibrated? The response curves vary for the ECM-8000. My ECM-8000 mic was sufficiently accurate without the calibration (after receiving the calibration curve), but I have seen several other calibration files for ECM-8000's that revealed that their calibration was essential.

2) That 4 dB peak in the 11 kHz to 17 kHz range is a little high for most people. Everyone varies in their upper frequency tuning preferences. Depending on the tweeter, I aim for flat to -2 dB in the 10 kHz to 15 kHz region, measured 10 to 15 degrees off-axis.

On another matter, have you considered contouring the top edge of the baffle the same as the sides? You have spend so much effort with good results on the front baffle shape to minimize baffle diffraction and baffle step. The front baffle top edge is fairly close to the tweeter, and not too far from the midrange.

:lol:
At this point my mic wasnt calibrated, and I used a cal file from Claudio Negro's site.
The tweeters output are tweaket by ear, and I thought it was may age who made this sounding good to me.
After calibration I found out that this measurement are (close to) correct below 4khz. Above 4khz my mic was peaking.
This plot is in fact +3dB@14khz!

And no, I havent consider any further baffle shaping.

[Paul W]

We aren't seeing off-axis measurements, woofer/mid xo etc, but you might have an off-axis "flare" just above xo. This would be caused by the narrow directivity of a large mid crossing to a broad pattern at the low end of the tweeter's range. Listening off-axis, or with the speakers close to walls, can make this worse. Sometimes a "fix" for the situation is a shallow BBC dip immediately above crossover.

If you want to flatten the peak in the TDFC, try a parallel RLC in series with the tweeter.

[norcad]

With the speakers a little offaxis, I didnt need the notchfilter on the tweet.
Here is the crossover used:

[norcad]

With this xo, the speakers stayed in the living room, for a couple of months, before my wife decided that they needed paint and some fabric at the front! 8O

The sound was good, but not great, and as this was my first speaker project, I was pretty sure that the drivers could do better.

I started to investigate what I could do, to make it better. First of all I was wondering about the peak/dip combo from 6-700hz and up.

To make a VERY long story short, it ended up with a much larger mid enclosure, with a lot of stuffing, I also tried a different mid, the Seas MCA15RCY.
And the conclusion was that it is the baffle shape that make this "problem".

After some hours playing with Baffle Diffraction Simulator, I decided to try something new.
Why not move the diffraction problem to the tweeter?
And Make a new front thats including the front fabric?

Well here we go:

New baffle:




And something I don't know what to call in English?




The result:

[TacoD]

I would expect that the wide baffle ensures enough mid output.

[Dave Bullet]

Hi Norcad,

By further increasing the Z (acoustic) centre offset by bringing the mid and tweeter forward, you will need to rework slopes to get phase alignment.

I also presume now you will be measuring on-axis to the mid not tweeter since you've swapped them over and the mid will be on-axis.

Not sure where your extra midrange enclosure space comes from. But the extra stuffing won't hurt.

David.

[Dennis H]

And something I dont know what to call in english?

That looks like a grill frame. Grill frame + grill cloth = grill.

I'm amazed at how good the English is of forum members from all over the world. You guys put me to shame with my (lack of) mastery of Spanish and German, my second and third languages.

PS, I just noticed that Google added Norwegian, Swedish, Danish and Finnish to their translation page. Now I can read all those cool Scandinavian speaker sites.

http://www.google.com/language_tools

[norcad]

Hi Norcad,

By further increasing the Z (acoustic) centre offset by bringing the mid and tweeter forward, you will need to rework slopes to get phase alignment.

I also presume now you will be measuring on-axis to the mid not tweeter since you've swapped them over and the mid will be on-axis.

Not sure where your extra midrange enclosure space comes from. But the extra stuffing won't hurt.

David.

I dont understand why the Z centre offset should be increased? Can you please explain.
Measured onaxis to the mid, the offset should be decreased, or?
Maybe you are thinking on the woofer, but that is not an issue, I'm using active xo and biamping between tweet/mid and woofer.

If you look at this picture, you'll se that I made a room for the xo behind the midrange. I just took that "wall" out, that increased the mid enclosure volum from 5,4 to ca 10L.

[norcad]

That looks like a grill frame. Grill frame + grill cloth = grill.

I'm amazed at how good the English is of forum members from all over the world. You guys put me to shame with my (lack of) mastery of Spanish and German, my second and third languages.

PS, I just noticed that Google added Norwegian, Swedish, Danish and Finnish to their translation page. Now I can read all those cool Scandinavian speaker sites.

http://www.google.com/language_tools

Frame was the word I was looking for! Thank you Dennis H!

Grill is what we call it in Norway to, so I should try to remember that

[Dennis H]

I dont understand why the Z centre offset should be increased? Can you please explain.
Measured onaxis to the mid, the offset should be decreased, or?

Doesn't look like a problem to me. You moved the mid and tweeter the same amount so the MT phase won't change. And a few mm aren't enough to bother the WM phase where the wavelengths are a meter or more.

[Dave Bullet]

Yup true (beware - a phrase you don't often see on boards follows) - my mistake.

David.