[ultrachrome]

I started this project back in '05. The goal was to design a hybrid 3-way with a highly efficient mid/tweeter section that I could drive with any amp project I might cook up.

I cobbled together a suprisingly nice sounding system with an uncalibrated ECM8000, TrueRTA, and a DCX2496 with Bill DiPaolo output boards.  But the complexity of living with an active three way system and being overwhelmed by SoundEasy along with life's other complexities led me to shelve the project.

I've finally picked it back up.  I just sent my ECM8000 to Kim G for calibration, ordered an upgrade for SoundEasy, and bought John K's ABC Dipole spreadsheet. I've decided to build a U-frame per ABC Dipole.  I'm finding John's documentation to be pretty accessible. 

Here are my drivers:
Fountek NeoPro5i  1wSPL=100dB
Audax PR170M0:  Fs=117  Qts=.51 Xmax=.5  1wSPL=100dB
2 x Lambda Dipole 12:  Fs=25.9 Qts=.68 Xmax=12 1wSPL=90dB

Crossover originally used was 250Hz and 2500Hz.

This is what I'll likely go with for my first test baffle.  I set the width to aid placement in my room and for a 300Hz crossover. Dimensions for the U-frame woofer enclosure are 8" deep, 13" wide, 25 3/4" high with a shelf between the woofers so they each have their own chamber.  These dimensions are with 1/2" thick walls and a 1" thick baffle.  Mid/tweeter baffle is 1" thick, 13" wide and ends 1" higher than the tweeter frame.  Test baffle will likely be MDF (have two sheets sitting in the garage) and final in baltic birch.  Thicker material and round overs being considered.


The wife actually wants a rectangular baffle without the U frame but I'd like a chance to get as much low extension as is practical and take advantage of the 6dB bump.  Currently I only have 100W for each woofer channel.

Here's my current baffle.  Larry Selmer talked me into this design.  There are no parallel sides.  I'm not necessarily going to abandon it but I'd like something a little more tried and true as a basis of comparison.



Am I missing anything?

[ttan98]

Hi,

I am currently thinking of building something similar to your system, the 2 bass drivers can be mounted in either U-frame, H-frame or just plain flat baffle. I think I need a simulation program similar to what Rudolf is doing in order to determine the ideal configuration.

Since you have built a prototype can please advise me what you think of the drivers particularly
1. Ribbon tweeter, is the back open?
2. Audax mid
3. AE bass drivers

Those who have use the Audax thought that they are very dynamic, but sound can be a bit thin,  a  lack of fuller sound. What is your verdict?

For ideas for making the frame please go to this site:

http://undefinition.googlepages.com/diy-sunflowers

mid-way down the page.

Since the diameter of the mid range is smaller than the bass driver the baffle width of the mid-range should be narrower than the bass driver to prevent a bump at the mid bass, as shown by some people here. With a narrow baffle it becomes more difficult to build the frame.

Cheers.

[Rudolf]

Here's my current baffle.  Larry Selmer talked me into this design.  There are no parallel sides.


Am I missing anything?

From my very personal point of view your new concept looks like a step back from what you already have. This may sound really arrogant, but I always found that Larry Selmer had given very good advise to you. I could even imagine how you could "radicalize" that design even more instead of going retro to more rectangular "boxiness".

How about letting the side wings end at the height of the Audax to minimize side reflections and the effective baffle size seen by the midrange driver and the Fountek? I would not cut off the existing side wings all from the bottom, but only start from the middle of the upper woofer with a gentle curve towards the main baffle.

In the same manner I would taper the top third of the main baffle - starting from where the side wings end. At the top of the Fountek the baffle should not be wider than the Fountek itself.

I'm not necessarily going to abandon it but I'd like something a little more tried and true as a basis of comparison.

May be that rectangular design is a little more tried - but your Selmer design is much more true. That´s what I believe.

Rudolf

[Saurav]

I've used the PR170M0 as an open baffle mid for a few years now. In my not-at-all-an-expert opinion, 250Hz is pushing it with that driver on an OB. You'll need a fair amount of boost from the active XO/EQ on a 13" baffle, and that driver only has 0.5mm excursion, so your peak SPL will be limited by that. ABCDipole has a section that lets you calculate what SPL your driver can produce on your baffle, run the Audax through that before you settle on a 250Hz XO.

FWIW, in my current speakers the Audax is crossed to an AE Dipole12 OB12 (corrected) (one per side) at 500Hz.



That's also why I used the OB12 instead of the IB12/15 - the 500Hz XO is easier with this driver. The Lambda Dipole12 should go higher more easily. You should also look into B&C and PHL - they have high efficiency mids with more excursion that may be good candidates for your project. That will probably be my next iteration, whenever I decide to try and make some changes.

[ultrachrome]

Rudolf, I think I see what you are saying.  I really appreciated Larry's help but he made my head spin.  I understood him at a high level but I quickly got overwhelmed.

If I taper the upper baffle, I move the dipole peak higher.  How do I deal with dipole roll off at a higher frequency?  EQ?  Higher crossover frequency?  Or both?

What are your thoughts on the U-frame?

[Rudolf]

ultrachrome,

I looked through John K´s explanation of A, B, C dipole on his website. His resulting midrange recommendation is almost exactly what I would have given as advice. Which actually is no surprise since I very much follow John K´s thinking about well designed dipoles.

Since you own both SoundEasy and A, B, C Dipole you can´t do better than take the baffle width and frequency ranges from ABC and do the crossovers with SoundEasy. ABC doesn´t give you different baffle width for midrange and tweeter. But it would always be good to make the tweeter baffle much narrower than the midrange one. You could test that by filling the tweeter values into the midrange chapter of John´s spreadsheet.

Looking closer John´s baffle width recommendation follows the "twice the cone diameter" rule. You should adopt this for the tweeter too. In many cases this will not be possible because the physical tweeter is much larger than its cone, but get as close as possible.

Doing a three-way-OB with John´s baffle calculations would result in three rectangular baffles put one above the other. I believe that it´s more elegant to join these baffles to an integrated one with diminishing width from bottom to top - as Larry did.

WRT U-frames my experience is an ant, John´s is an elephant. So why should anyone ask me?

Rudolf

[tg3]

You should also look into B&C and PHL - they have high efficiency mids with more excursion that may be good candidates for your project. That will probably be my next iteration, whenever I decide to try and make some changes.

PHL 1120 +1

[Rudolf]

ultrachrome,
for your PR 170 my simulations give a 10" wide baffle if you cross over at 1 kHz. As I see it the Fountek would like 2 kHz better. In that case the baffle for the PR 170 should be 7" wide. The larger baffle will be about 2 dB more efficient from 200-500 Hz.
As Saurav mentioned you will have to take your bass drivers as high in frequency as ever possible. For that reason I see your U-frame no deeper than 5-6" inside.

[pedroskova]

edit...delete...need more coffee.

[ultrachrome]

How would you envision a 7" baffle for a 7.5" driver frame?  Let the frame overhang the edge?  Rectangle or trapezoid?  Baffle thickness?

Does a roundover contribute to the baffle width with respect to dispersion?

I'm still trying to mentally picture the relationship between baffle width and dispersion.

[Rudolf]

How would you envision a 7" baffle for a 7.5" driver frame? Let the frame overhang the edge? 

Ouch! I did not look for that. But baffle width is not like BR vents. You don´t need to be exactly spot on. Simply make the baffle as wide as the driver frame.

Rectangle or trapezoid?  Baffle thickness?

I´ve never seen any research regarding different baffle shapes except circles and rectangles. EDGE would be perfect to get a feeling yourself.
WRT baffle thickness: As with almost every question in loudspeakers you need to see dimensions in relation to the wavelength you are looking at. 1" difference at 1 kHz? Do as you want. 1" difference at 10 kHz? It´s another world!

Does a roundover contribute to the baffle width with respect to dispersion?

Same answer as above.

I'm still trying to mentally picture the relationship between baffle width and dispersion.

Did you have a look at this:http://www.audiocircle.com/index.php?topic=64048.msg585002#msg585002
If yes, what is your problem in particular? Is it some formula or something visual?

[ultrachrome]

I think I'm understanding it more especially after figuring out how to simulate off-axis response in Edge.

The simulation of a trapezoid midrange baffle with an average width roughly equal to the driver diameter, shows a much smoother off-axis response than a 13" rectangular baffle.

But I'm still confused.  What's the goal, smooth off-axis response or off-axis bandwidth?  By reducing the midrange baffle size, I'm not getting any more high-frequency extension using.

[Rudolf]

But I'm still confused.  What's the goal, smooth off-axis response or off-axis bandwidth?  By reducing the midrange baffle size, I'm not getting any more high-frequency extension using.

The goal clearly is smooth off-axis response or - more focused - power response. And yes - you have to trade in bandwidth for it. But off-axis bandwidth is limited by two factors: Baffle width AND cone diameter. If your speaker is already beaming, reducing baffle width is useless. That´s one reason for the "baffle width should not excess twice the cone diameter" rule. It´s valid the other way too: Cone diameter should not excess twice the baffle width! If your baffle needs to be smaller then you need to look for a smaller driver too.

As good as the Audax may be in itself - for dipole use it is heavily compromised: for a low crossed midrange it´s Xlin is too low, and for a high crossed mid it´s cone is too wide. Try reducing the speaker size in your EDGE model and you should get more high-frequency extension off-axis until the baffle starts to be larger than twice the cone diameter again.

[ultrachrome]

But I'm still confused.  What's the goal, smooth off-axis response or off-axis bandwidth?  By reducing the midrange baffle size, I'm not getting any more high-frequency extension using.

The goal clearly is smooth off-axis response or - more focused - power response. And yes - you have to trade in bandwidth for it. But off-axis bandwidth is limited by two factors: Baffle width AND cone diameter. If your speaker is already beaming, reducing baffle width is useless. That´s one reason for the "baffle width should not excess twice the cone diameter" rule. It´s valid the other way too: Cone diameter should not excess twice the baffle width! If your baffle needs to be smaller then you need to look for a smaller driver too.

As good as the Audax may be in itself - for dipole use it is heavily compromised: for a low crossed midrange it´s Xlin is too low, and for a high crossed mid it´s cone is too wide. Try reducing the speaker size in your EDGE model and you should get more high-frequency extension off-axis until the baffle starts to be larger than twice the cone diameter again.

Thanks.  So I'm seeing the trade-offs now.  Insisting on high-efficiency drivers means that I'm not likely to find a smaller diameter mid.

I think I can deal with the low end limits of the PR170M0 but I may have to live with some beaming on the high-end.

[ultrachrome]

That's also why I used the OB12 instead of the IB12/15 - the 500Hz XO is easier with this driver. The Lambda Dipole12 should go higher more easily. You should also look into B&C and PHL - they have high efficiency mids with more excursion that may be good candidates for your project. That will probably be my next iteration, whenever I decide to try and make some changes.

If I'm using the spreadsheet correctly, Xmax limited, I reach 110dB at about 380Hz.  Max XO is 580.  So you're pick pretty much in the middle.

At 45 degrees, the Dipole12's start rolling off just above that point.

[Rudolf]

So I'm seeing the trade-offs now.  ... I think I can deal with the low end limits of the PR170M0 but I may have to live with some beaming on the high-end.

You can´t have it all for free. To get baffle width and driver size fully optimised from 20-20000 Hz you probably would need a four-way system. Think about driver sizes of 0.75" - 2.5" - 7.5" - 21". That´s a 3-fold step from driver to driver. In real life you would resort to a 3-way system, needing a 4-fold step: 1" - 4" - 16" (12+12). You would forget about getting the lowest and highest octave right and still need to compromise a bit in between.
But you will still be lightyears ahead of that single driver in one rectangular baffle. 

[Saurav]

Could you explain this further?

As good as the Audax may be in itself - for dipole use it is heavily compromised: for a low crossed midrange it´s Xlin is too low, and for a high crossed mid it´s cone is too wide.

Are you talking about the cross to the woofer in one case and the tweeter in the other, or the woofer crossover in both cases?

[Rudolf]

As good as the Audax may be in itself - for dipole use it is heavily compromised: for a low crossed midrange it´s Xlin is too low, and for a high crossed mid it´s cone is too wide.

Are you talking about the cross to the woofer in one case and the tweeter in the other, or the woofer crossover in both cases?

Woofer in one case and the tweeter in the other.

In a dipole system a drivers low end is almost always challenged more than in a CB or BR. So at a given diameter you need more excursion. The ORION midrange has 10 mm, the NAO midrange drivers 5,5-6,5 mm. This illustrates how far off the Audax is.
BTW: The 7" midrange driver in my own dipole speaker has 5,5 mm.

[ultrachrome]

BTW: The 7" midrange driver in my own dipole speaker has 5,5 mm.

What is it?

[Saurav]

Thanks. Assuming I did the math right back when I built my speakers, I figured I'd be limited by the 3.5W output of my SET amp before I'd hit the excursion limit on my Audax, especially given my high-ish 500Hz XO frequency.

But I agree with you, and if I were starting from scratch I'd use a different midrange driver. I only used this one because I already had it.