New project: Usher 8137a, T-B W3-1335SB, SEAS 27TAFN. WMTM, open baffle mids.

[fjhuerta]

Hello all,

I've officially begun construction of a new 3 way speaker, still to be named.

Since the front baffle will be painted matte red, and the enclosure is going to be made out of Banack or a similar dark wood, I'm leaning toward "The Meteors". That, and my son is a big fan of Speed Racer now, and his name is "Meteor" in spanish. We'll see.

How far am I on this project, to dare call it "non-vaporware"?

* The drivers are all here.
* I've already built the baffle and cutouts.
* The enclosure should be ready and finished early next week (I'm having it made by a professional furniture maker).

So, I guess it'll be completed and voiced in two months, tops.

Here are the details of my design.

* The speaker will be a clone of Jed's Tombstones, but with different drivers. I thought about using the Dayton RS270, RS150 and RS28a (which I also have), but I decided to start with a smaller version.
* It's not that small, though. External dimensions are 1.10 meters tall, 42 centimeters wide and 47 cms deep. The 8137a apparently likes a big enclosure.
* The mids are the Tang Band W3-1335SB in an MTM configuration. I chose this driver because I like the w4-1337 drivers, and I thought I could use a smaller one in an MTM to minimize CTC spacing. Indeed, the size of the MTM cluster is extremely small - barely 26 cms. long.
* The tweeter is one of the smallest metal domes around - the SEAS 27TAFN.

I'll post pics next week of the enclosure. Preliminary design ideas are:

* Duelund crossover?
* Crossover points around 500~600 Hz and 4 KHz.
* I'd like to see if I can get away with 2nd order curves (if so, then the Duelund is a given). If not, 4th order L-R curves will be used.

[JonMarsh]

Sounds interesting, Javier! How do the W3-1335SB test out in the low end?

[fjhuerta]

Hi Jon,

I haven't done any distortion testing yet. I don't expect much out of them, with 0.5 mm of Xmax.

I'll do the testing early next week - FR in-baffle and HD sweeps!

[fjhuerta]

Question - can I do HD sweeps on an OB, or do I need to build a box?

[JonMarsh]

Yes, you can do HD sweeps that way- I've tested sub drivers that way, and others. It's best to do it fairly near field, to minimize the baffle influence on the measured response.

You can do HD sweeps even on open free standing drivers, but then you must work very nearfield, and be sure not to overload your mic or electronics.

[fjhuerta]

Thanks Jon,

I'm not feeling very confident right now. I ran an informal impedance sweep on the W3's and there's some sort of irregularity around 1.3 KHz.

I'll publish it as soon as I get back my laptop, but it doesn't look very encouraging at all.

[---k---]

Sounds like an interesting project. Good luck.

For a name, with the way you were describing it, I thought you were going to say that you were leaning toward Matador. I think it would be another good fit.

[fjhuerta]

Matador sounds great...!

I'm still worried about the impedance of the W3's. I'll post it later so you can check it out.

[fjhuerta]

Some progress has been made...

I have the cabinets!

I reinforced them today. I added damping via foam. Installed the drivers. Everything is ready for testing, sometime this week.

The baffles are very wide. I wanted to try a wide baffle design. At 42 cms., it's a compromise between baffle size and looks. I think it looks a bit too wide, though - I had to do it in order to reach 90 liters of volume.

[fjhuerta]

And, some measurements I managed to take - free air.

Usher 8137a



Seas 27TAFN



One Tang Band W3-1335SB



Two Tang Band W3-1335SB



I notice an impedance irregularity on the Tang Band W3-1335SB at around 1.4 KHz. I hope it isn't too serious on the FR plot.

[Johnloudb]

Looks really nice!! I'm really interested in hearing your comments about the woofers. I've got six of them I'm going to use in a dipole speaker. I have High expectations for them .... hope I'm not disappointed.

[JonMarsh]

First guess about that impedance irregularity is a surround resonance, at that frequency. Proof is in the pudding, though- everything else is speculation.

[fjhuerta]

John,

I do, too! For the price, they look seriously well built. Also, they look pretty cool - much better than in pictures.

Jon,

I'm still learning about these kinds of things, but one thing I noticed is that the dustcap on these little drivers is 3/4 of the cone... could this be part of the problem?

Anyway, I hope to do ASAP the FR and HD sweeps.

[fjhuerta]

I found this review of the W3 on Parts Express. He does mention the 1 KHz anomaly (I suppose that, when combined with the baffle shape, that's the frequency where the problem is).

Paul Wilke from The Netherlands


This is a very well built little driver that is able to handle a lot of power and produce good output relative to its size. I have been able to get a -3dB at 70 Hz with an oversized enclosure (3.3 litre, about 3 times Vas). It needs taming though. In my enclosure I use a mild notch centred at 1Khz (2.2 mH, 15 uF and 3.3 Ohms in parallel) and a hard one at 15KHz (small capacitor over 1 mH crossover inductor) to suppress an aggressive break up peak. Because of this break up node, it is not possible to use this driver as a full range. Complete crossover assembly with two coils adds about 3 Ohms in series with the driver, which is welcome to lower the Qt for use in this oversized cabinet. Combines very well with Dayton ND20FB tweeter.

If you get the crossover right, the reward is a very clean and accurate extended midrange and, for a speaker this size, remarkeable bass.

Because of the complexities of controlling the behaviour of this driver, I would recommend it only to folks that have the necessary measuring equipment and the skills to compensate for its quirks.

[fjhuerta]

This is a graph I took a couple of months ago of the W4 vs. W3 titanium drivers.

At the time, I thought it wasn't representative of the behaviour of the W3, since it wasn't flush mounted, and the baffle wasn't very well built.

Hopefully, this is true. Because the anomaly at 1~2 KHz is there...

[fjhuerta]

Good news (at least for me). The W3 doesn't seem to be a bad driver after all.

Nearfield:



Farfield, single driver on the OB, on-axis. The dip should smooth out when both drivers are used.



Farfield, single driver on the OB, off-axis.



Harmonic distortion (I need help interpreting this).

[fjhuerta]

Some sims.

The FR of a single mid driver on-axis has a huge dip. I believe Jed's Tombstone design helps a lot in getting rid of it. The closer the upper driver is to the baffle border, the best FR looks.

Here's the simulation: Single driver on axis (just like my measurement).



Compare to both midranges, with the listening axis on tweeter axis.

[TacoD]

You put a lot of effort in these speakers. What do you think of the Usher woofers? When I used those, I was impressed how these performed in a closed box.

edit: I also used a pair in a vented cabinet, a true bargain for the performance you get.

[BOBinGA]

Javier,
Have you been able to get the Edge to agree with your actual measured response? I haven't. I built an open baffle and modeled an RS150S in the Edge and got this:



When I measured the driver in the baffle, I got this (the top dark blue line):



Disregard the aluminum cone ringing at 10K hz. The light purple line is my first attempt to flatten out the hump at 1000 hz and the ringing at 10K hz with the crossover. But I must be doing something wrong in the Edge since I'm getting no agreement at all. I just decided to use the actual measured response and go from there. The Edge doesn't seem to help.

-Bob

[fjhuerta]

Hi Bob,

I only use Edge for getting an overall idea of how FR may look. For example, the FR dip at 1.5 KHz was expected, although it happened at closer to 1,250 Hz when I measured it. I always trust measurements over simulations.

I want to measure both midranges now, to see if indeed, the dip is smaller.

Also, the graph you posted has a 30 dB range. Every imperfection will look huge! Mine were taken with a 60 dB range.

For now, I can say The Edge really helped me regarding placing the drivers. So far so good...

[Mazeroth]

Bob, how were your measurements done? Equipment, distance, axis, was any software used? No offense, but it looks like you used a Radio Shack SPL meter and a spreadsheet :B

[BOBinGA]

Makeroth,
You are right. I did use a RS meter and a spread sheet. And no offense taken, that's all I have. But I have a pretty accurate compensation curve, so the chart is more than in the ballpark. Plus the peak is in the very flat part of meter's response, and it matched my listening experience.

The measurements were take at about 10' farfield. I took the same measurement on axis from a couple of locations in the room and they were consistant. I don't want to hijack Javier's thread so I'll post some pictures later in another thread, but the point is that I could not get Edge to accurately predict what I heard. I played with it for a long time to choose the location of the speaker on the baffle so that I got a reponse that was flat to within plus or minus a couple of db then built the baffle. What I ended up with was no where near what I got with Edge. Oh well, I got the peak fixed now and it's on its way to being finished.

Javier,
Keep up the good work on these. It's good to see a new original project. These are similar to what I'm building, so I am very interested in your results.

-Bob

[fjhuerta]

Thanks Bob. I'd have to say that using the Radio Shack meter isn't the most accurate way to do measurements. That's probably part of your problem.

I've been working at my speakers - its just that I've been doing stuff that needs to be done, yet doesn't help during the design (building grilles, wiring all drivers, soldering connections, etc.). Hopefully by this weekend I'll know exactly how everything will behave.

TacoD, have you seen John's measurements on the Usher woofer? They look very fine up to around 1 KHz. I can't wait to see how they'll measure in my box (90L, vented) - since the wide baffle apparently smooths out the hump below the baffle step. I think the crossover for the woofer will be extremely easy (I'm not so sure about the midrange, though).

[TacoD]

Yup I have seen those measurements, the HobbyHifi magazine also published some distortion graphs (very low distortion). The design of HobbyHifi used the driver in 100L vented with very low tuning.

Image not available

more info on http://www.audax-speaker.de/content.php?session=7872c514e18aa9ebe6af057edf520a c5&seite=shop/produkte.php&details=56&hauptrubrik=2&lang=en

[fjhuerta]

Those look nice.

My box will be 90L too, tuned to around 28 Hz. The woofer really loves a big enclosure when using ports.

[fjhuerta]

Simulations were right on the money for the baffle!

The W3-1335sb's are stunningly flat from 1.5 KHz to 7 KHz. Below that, the open baffle causes a hump from 1 KHz to 250 Hz -roughly 5 dB. Much better than before.

I took 36 measurements for this speaker (on axis, 15°, -15°, -30° for each driver, with rear mesh, without meshes, and with front and rear meshes). Every measurement seems to indicate the same - the W3's and the Seas tweeter are going to be a great match.

I'll post my progress tomorrow night.

[dsrviola]

A friend and I are building .....

....a set of HT L+R speakers centered around those TB's. I love them so far (I've just listened to one channel with a prototype crossover)

Here's a link to my friend's website that tells a little about the project:

http://www.lonesaguaro.com/speakers/XavHT/XavHT.html

More once I have the speakers assembled and installed. I can tell they're already a huge leap ahead of my previous HT L+R ( X158 ):

http://www.lonesaguaro.com/speakers/150MTM/RS150MTM.html

[fjhuerta]

I already finished the crossover. The extended FR of the Tang Bands wasn't really useful here, since I'm running them up to 3.4 KHz.

I tried a "standard" 4th order acoustical. It did wonders to the on-axis response, but off-axis greatly suffers. When simulating them at 30°, the diffraction hump (I suppose mostly related to the dipole configuration and baffle size ) caused a big hump at 1 KHz.

The Duelund configuration helped greatly in this regard. The hump is still a bit there, but far more controlled and gentler looking.

After looking at Stereophile's measurements for the Jamo open baffle loudspeakers, I'm still wondering wheter to supress the midrange a bit. I'll be ordering the crossover parts today or tomorrow, and I'll be sure to have enough to modify the midrange response and shelve it down, if necessary.

My findings will be posted later today (hopefully!).

[BOBinGA]

I got a big hump at 1200 on my dipoles also (see my graph above). It sounded terrible, like the singer was using a megaphone instead of a microphone I had to use a series trap to take out the hump and it sounds very good now. I think you will find you need the extra parts, so add them to your order.

[fjhuerta]

Actually, the Duelund topology took care of the hump at 0, 15 and 30°. But I suppose that at greater angles problems may still be there.

[fjhuerta]

Finished!

I may never go back to an enclosure... ever. At least for the mids. ;x( I've seen the light, and it's bright, open, airy and... wow.

I'll need quite some time to post about how it got to where it is right now. Suffice to say that:

1) Desigining an open back speaker is counter-intuitive. Software tells you "it's not optimal, you doofus!". RTA measurements tell you something is very, very badly designed. And your ears are like "huh?".
2) I would NOT be surprised to learn that this particular design only works correctly in my room. As a matter of fact, I'm pretty sure that's the way it'll be.
3) The Tang Band mids are my new favorite midrange driver. Yep, I like them more than the W4-1337s or the Dayton RS-52s. Or, it could be that open baffle really lets them sing.
4) This all started as a Duelund and ended up... well, I'm not really sure.

It'll all make sense (hopefully) soon enough.

Meanwhile, here's an RTA measurement I took on my listening position. Red is left channel, blue is right channel. The speakers measure identically - it's just that I've found out (the hard way) that open baffle speakers react to their surroundings in new (to me, anyway) and exciting (yeah, right) ways.

[Johnloudb]

Well, that looks really flat to my eyes. Don't know why you're complaining about measurements, but then again I don't know a whole lot about measuring loudspeakers. I guess all that bass boost is just room gain? Anyway glad your happy with them.

So how do you like the Usher woofer?

[fjhuerta]

Yeah, bass boost is room gain. The graphs are flat because I hammered the RTA curve into something decent. You'll see that things started looking really terrible with V1.0 of the crossover - the one that JustMLS and LSPCad thought was "awesome".

In this regard, I learned a lot from Jim Holtz's Statements (http://www.geocities.com/cc00541/Statements.html) and the Jamo R907 speaker (http://www.stereophile.com/floorloud...r/index4.html). As a matter of fact, the Jamo's measurements on the site helped me a lot with the final crossover tuning.

As for the Usher woofer - it's awesome. So easy to work with, and sounds great in this application. Bass is far better than my dual RS-180 design. Then again, the woofer is tuned at 28 Hz on a 100L box, so it's sort of a little beast. :T

[Johnloudb]

I'm glad to hear that about the 8" woofer. I have high hopes for a speaker I'll be building later this summer that uses that Usher. Glad all your work on getting the response flat paid off. :T Maybe, I'll be so fortunate.

[fjhuerta]

Measurements

First of all:

Measurements: 1.8 M, on tweeter axis, 6 / 10 ms. I'm only including a single woofer measurement at 0° since off axis it behaves like a typical 8" would.

The interesting thing here is the midrange. The more you go off-axis, the bigger the hump centered at around 500 Hz gets. Nice, free gain you get when going open baffle - which will set the lower frequency limit for the crossover. Well, that and the FR, which drops so fast below that you don't have much to work with if using 2nd order slopes, as I did.

Tweeter, 0°



Tweeter, 15°



Tweeter, 30°



Midrange, 0°



Midrange, 30°



Little did I know (well, I had the basic idea) - those midrange measurements would come back and haunt me when a back wall was added to the mix. But I'm getting ahead of myself.

[fjhuerta]

Here's the woofer response at zero degrees, with a 10 ms gate.



I took a lot more measurements - including a set with rear meshes on, front meshes on, and both meshes on, at 0, 15, 30 degrees. Rear meshes on didn't affect the measurements; front meshes subdued the top end somewhat and added a measure of diffraction. So I modeled using rear meshes - no meshes, which were identical.

So far, so good. Now I needed a crossover topology. I settled on a Duelund filter, with a center frequency of 1,300 Hz and an Aleph of 3. Then designed a circuit around it.

And things looked great.

Zero degrees.



15 degrees, horizontal.



30 degrees, horizontal.



At this point, I thought the 500 Hz hump would be my biggest problem, but I hoped that power response and appropiate positioning of the speaker would take care of this. The further you go along the horizontal axis, the bigger the hump. I found that taking real measurements off axis really helped with this design, and I'll definitely be using them from now on (LSPCad's simulations were completely off-base here).

Here's the crossover circuit I used to achieve this response.



Nothing out of the ordinary - simple 2nd and 1st order circuits all around, with some shaping components on the woofer and midrange. Also, there's a shunt on the woofer to shape the higher frequency response, and LRC traps on the midrange and tweeter, flattening their resonance hump. This is especially important on the mid - if I didn't use it, I'd get a big FR hump... like this.



This is one of Jamo's open baffle speakers, the R 907. John Atkinson's measurements were invaluable for designing this speaker, actually. Notice the lower end of the midrange - there's a sharp peak. I believe no impedance compensation was used on the Jamo, because I had the same issue until I added the LRC trap.

The crossover has several elements in parallel for power handling, and to reach certain values. I used iron core inductors for the shunts, a Jantzen P-Core inductor on the woofer and a mix of 20 / 18 GA inductors everywhere else. Shunt caps are electrolytics, and the rest are Dayton Polys.

(The sound you hear is Evil Twin about to murder this young Padawan for using substandard components on this design but hey - this is a recession. I'm acting accordingly!).

[fjhuerta]

While all this happened, I wondered how Jamo could go from this...



(average 30° 1 meter FR)

To this.

1/6th spatially averaged response in listening position, with / without grilles.

I don't know!



I guess you do, and I did in a way, too. It's just that I've never been too comfortable designing *against* what LSPCad thinks its optimal. This would be a personal challenge for me - how to design a speaker while ignoring some suggestions LSPCad makes.

Anyway,

Here's the relevant info for the crossover I posted above.

Impedance:



Transfer function:



Time domain:

[fjhuerta]

Listening impression

Well, the speakers didn't blow up the first time I listened to them.

First thing I noticed - the gorgeous, GORGEOUS midrange. Seductive. Silky. An amount of textures I rarely hear on speakers.

Second thing I noticed - well, something IS wrong here. There's too much detail.

Third thing - yep. There's some extra midrange energy at work here.

It was time to take out the RTA. The part I feared the most - up to this point, LSPCad had guided me through the design project. From now on, it was trial & error, intuition and (my lack of) experience.

[fjhuerta]

So, what was wrong?

Well, this.



Black is the RTA measurement of my right channel, vs. the measured response at 1.8M.

Time for my ultra-high performance passive FR shaping device.



I'm not including those RTA measurements, but you get the idea. It's quite obvious - the rear of the drivers' output combines at a certain frequency (I suppose, depending on your wall material, the distance to it) and emphasizes certain parts of the spectrum. In this case... well, you can see what happened. It doesn't look nice, and it sounds even worse.

Fortunately, the midrange circuit is so simple, you can shape the response by varying the first resistor in the circuit.

Unfortunately for me, it took a couple of days to find out about it (I changed every conceivable piece of the network). But hey -knowledge hurts, doesn't it?

So, what did I end up with?

Remember.... this?



Well, how about... this?



This change makes the speaker sound... like... this.



With a simple change: double the value of the midrange resistor. I actually made it 12 Ohm because I had a 20 and a 30 available.

[fjhuerta]

How do they sound...?

Let's get this out of the way...
"Best speakers I've heard ever... best bang for the buck... veils have been lifted... silky smooth midrange... thundering and precise lows... blahblahblah".

Seriously... this is what I think about them.

• They sound big. No, make that HUGE. If I close my eyes, I can't imagine the size of the speakers - because they sound so big. This is the first thing that caught my attention - after all, they aren't really that big (3" woofers, 1" tweeter, 8" woofer).
• They, as so many people have said before, don't do pinpoint imaging well. This is due to two things, IMHO. First, I built a "Wide Baffle" speaker (it's almost 45 cms wide). Second, it's open baffle. They are amazing for creating a wide, deep soundstage, but they don't do precise positioning of instruments. And I find the sound to be more natural this way.
• The midrange is incredible. There's far more texture and detail than in the boxed Neo8 I used. Not having an enclosure really does great things.

I guess I'll add more to this thread as days go by, but I'm very excited about this. I didn't know how good open baffle could sound. Then again, it has issues of its own - specifically, the extra rear energy. I wonder how different this design would sound with absorption behind them. I don't know.

I'm hooked now. Open baffle rules!

[JonMarsh]

A very interesting journey, Javier- and thank you so much for sharing your details from point A to point Z! I ran into some similar experiences working on my Arvo Part project years ago, but didn't have any kind hints from the Jamo speakers and Stereophile's review- you did a very nice job connecting the dots and coming to the conclusion (for now?) of your journey.

Now imagine the complications if your woofer is dipole, too.... :W

[Saurav]

Very interesting. What is the resolution of your RTA, and did you take the RTA measurement from the same mic location as the gated / MLS measurement?

[fjhuerta]

Jon, thanks for your kind comments. I forgot about the Arvo Pärt. I do remember the speaker being perfected all the time. Is there documentation regarding how you solved the extra energy issue? I wonder if an LRC parallel circuit would have been a better solution, but to be honest, the extra resistor does the trick so well I may not try using the trap.

Saurav, I used a 1/6th octave resolution. I did try going higher, but I found out I was concentrating in very small sections instead of "the big picture", so to speak. No, I didn't take both measurements from the same positions. The gated ones were taken at 1.8 meters, on tweeter axis. The RTA ones are from my listening position.

[BOBinGA]

Ha! You got a big hump at 1500 hz in your RTA measurements. Remember my post #19 saying that I got a big hump at 1200 on my dipoles that I had to beat down with the crossover? And you guys thought it was because of my inadequate measurement equipment. Now I admit my equipment is weak, but it did show the peak and my ears knew my measurements were correct.

(Sorry, I just had to get in an "I told you so" )

But seriously, This is a great project and I'm looking forward to the rest of your build description. ;x( It's good to have another mid dipole speaker to add to Paul's Aethers and Sunflowers. I haven't heard the Statements yet, but I'm sure the open mid is a big part of the reason they are so popular. And let's not forget Jed's Tombstones and Dan's Basslines. Word is getting out that dipole is better.


I have similar observations about dipole imaging being huge, but slightly less pinpoint than box speakers. But the clarity you get is addicting. The illusion of a full orchestra spread out behind your speakers is unmatched in the box speakers I've heard.

-Bob

[---k---]

Javier,

Looking good.

I remember reading once a comment (I think) by Paul (Undefinition) about how it critical it was to get the spacing of the speaker away from the wall just right. That he had to spend quite a bit of time moving the speaker back and forth to get the front wave and back waves aligned right. Have you moved the speakers around at all as you are taking RTA?