[MJK]

Rudolf,

With respect to your first plot, if you moved the driver off center in the 25 cm wide baffle I bet the dip would be greatly reduced. For a 3" driver which needs to be crossed over to a woofer anyway, why wouldn't you use the 40 cm wide baffle and design a crossover to roll off the low end at a selected frequency? Why would you want to require both EQ and a crossover to make the 11.5 cm baffle work?

With respect to the second set of plots, I am assuming the nulls above 5 kHz are due to lobing caused by the driver diameter. I am not really concerned about the response 60 degrees off axis, so looking at the on axis and 30 degree off axis curves the differences between the side by side plots are 1 or 2 dB. In my opinion this is in the mud, it will be minimal compared to peaks and dips caused by floor bounce, crossover integration, position of the woofer relative to the 3" driver, and room interactions.

And the final EDGE plot, the driver is in the center of a circular baffle so I am not sure why this is important to show. Nobody would use that configuration. If they used that set-up they will get what they deserve.

Please excuse me for saying this, but it seems you are making selecting a width during the design of an OB much more difficult then required. There are bigger hurdles to be addressed. I just don't see any benefit to going with a baffle that is so narrow that it requires both EQ and a crossover to make it work like the wider baffle that can use a simple crossover. I see many DIYers get too focused on perceived problems and expend significant energy to address something that will be lost in the bigger picture that is the final in-room SPL response. I am sorry but I still just don't get the advantage of a narrow baffle on the order of two times the driver diameter, maybe I am too dense.

Martin

[ultrachrome]

Not to throw fuel on the fire but I did a comparison of a rectangle v. trapezoid on-axis with a PR170M0.

The baffle is 13" wide and 42" tall.  The midrange is centered at 30".  At 25.5" the baffle tapers to 6.375" wide.  The baffles are elevated high enough that the middle of the midrange is about 1.2M off the ground.  I fashioned some cardboard to tape to the back of the baffle to make it 13" wide all the way to the top.

With a 4ms window and a mic distance of 1m I get identical output between rectangle and trapezoid shapes but the rectangle response is smoother noticeably smoother.  The trapezoid has a wider hump and a deeper, wider dipfrom around 400 to 3k.  Below 400Hz the rectangle has 1dB more output.  Otherwise very similar.

With the microphone at 0m, about 2cm away from the dustcap and the window enlarged to 200Hz, the rectangle has 2.5dB more output across the board.

I did not have time to take off-axis measurements.

I can post some screen shots if there is interest.

[Rudolf]

With respect to your first plot, if you moved the driver off center in the 25 cm wide baffle I bet the dip would be greatly reduced.

Sure, but off-axis response would be worse.

For a 3" driver which needs to be crossed over to a woofer anyway, why wouldn't you use the 40 cm wide baffle and design a crossover to roll off the low end at a selected frequency? Why would you want to require both EQ and a crossover to make the 11.5 cm baffle work?

It´s mainly about better localization:
We get localization cues by interaural phase difference (IPD) and interaural level difference (ILD). Below 800 Hz IPD is the dominating cue. When wavelength become the same magnitude as our head dimensions, phase cues become less significant. So at 800 Hz ILD starts to become relevant. At 1600 Hz ILD is the dominating cue. Loudspeakers with better controlled directivity above 800 Hz enhance the precision of source localisation. I noticed that when cutting off the sides of my baffles. It may not show if you are exactly in the sweet spot, but the stereo image keeps more solid to the sides.
40 cm wide baffles don´t control directivity above 800 Hz. I believe the diagrams show that.
I would not say that CD is a primary goal, but if you´ve got the response flat and the crossovers right, it´s a worthwhile next target.

With respect to the second set of plots, I am assuming the nulls above 5 kHz are due to lobing caused by the driver diameter.

Exactly.

And the final EDGE plot, the driver is in the center of a circular baffle so I am not sure why this is important to show.

I already had that illustration at hand. I know that it does not show good praxis, but it clearly illustrates the different "faces" of a dipole below and above the first hump IMHO.

Please excuse me for saying this, but it seems you are making selecting a width during the design of an OB much more difficult then required. ... I see many DIYers get too focused on perceived problems and expend significant energy to address something that will be lost in the bigger picture that is the final in-room SPL response. I am sorry but I still just don't get the advantage of a narrow baffle on the order of two times the driver diameter.

I am under the impression that both Linkwitz and Kreskovsy are protagonists of narrow baffles. In fact this "twice cone diameter" is not my invention, but has been hiding in Johns Tech Studies http://www.musicanddesign.com/Dipoles_and_open_baffles.html for quite a while.
That people sometimes are concentrating on perceived problems instead of real ones is part of human nature. So it´s never wrong to focus us back on what is really important. But then we should have to talk about the crisis ...

Rudolf

[giamba76]

thanks guys,

your systems are not that simple i see.
electronic crossover, active Xover.... biamped...

well i might buy something like that to, (like the berhinger)
the advantage is that you can always tune it if you like, (otherwise there isn't ? or is there ?)

so you split the signal in an electronic crossover between sub and fullranger
then it would be best to flatten the bass as good as possible (like the t1951)

i suppose you don't use an digital crossover (like dec2496) because the quality loss in conversion  ?)

but if you have 2 amps behind the crossover: how do you manipulate the volume or do you do this before the crossover ?

thanks

Yes
i split the signal into El.Cross. then i set (and i don't touch anymore) the tube amp volume in a way that the master volume on cx2310 and on T1951 (one for each ch) are set on minimum for a pleasure listening level, it's not so complicate to use, but in this way you have left and right volume knobs!!
I don't use DCX2496 for the reason you said...quality is not so good in conversion (for all: there is a french site where is well explain how to improve the dcx...)

[fillemon]

well guys, this tread is becoming a "tutorial"   

learned a lot from it. still learning, but most of all: i should not deviate from a decent design
because i don't know enough, and have not the knowledge and tools to test a design.

i understand that we should focus on the important things. i know that when you start measuring
a speaker things can look horrible, (at first i thought that, but then after seeing enough of those
i realised that flat curve over the entire room is absolutely unattainable).

i'm gonna read a bit more and then decide what to do.
thanx everyone

[ultrachrome]

It can be very overwhelming at times.  A lot of that is generally of our own doing.  Even folks trying to help often make it worse.

If you have funds to justify building an established design, that would serve you well as a reference for further research into crossover and baffle evolution.

Here's where I should shut up but I can't help myself...

It seems to me that with some basic maths, we could help adapt MJK's fostex based passive design to work with your fe167e to give you something decent to start with.  This assuming you're willing to adopt MJK's woofer and baffle choice and crossover point.

fe103e  Qts=.35 SPL=89
fe167e  Qts=.33 SPL=94

Pad the efficiency down 5dB and adjust the HP XO values to deal with a lower impedance at 500Hz, according to the spec sheets.  You could measure the impedance at 500Hz with a DMM.

Does it have to be more difficult than that?

[MJK]

If you want to combine the Fostex FE-167E with the Eminence Alpha 15A you might consider this design that I worked out last year and have posted in a few places.

http://www.quarter-wave.com/OB_3_Drivers_2_25_08.pdf

Basically it is a 20" wide by 40" tall flat OB with two Eminence Alpha 15A woofers and one Fostex FE-167E full range driver. Crossover is 200 Hz 2nd order low pass LR for the woofers wired in parallel and 500 Hz 2nd order high pass LR for the full range driver. The response SPL is about 95 dB/W/m from 40 Hz (-3 dB point) to the upper reaches of the FE-167E. To calculate the crossover component values assume the driver impedance is simply Re (remember the Alphas are in parallel) and use text book formulas, that should get you very close.

I have not built this system but the design philosophy is the same as my FE-103E and Alpha 15A OB speaker system which people have built and been very happy with the results.

[fillemon]

thanks people for your kindly reply.

i'm liking the idea of a dual alpha (or maybe a dual goldwood) because i get much more options when it comes down to amps for a 95db versus a 91db efficient speaker. it's a shame to tame down an efficient driver with a resistor.

at this moment i'm gonna try to get as much out of my fostex (in mltl) as possible. i have seen different mods: varnish on the cone, and a phase plug. are there any other and if i do these would it compromise the driver's use in an open baffle ?

another question: if the fostex is only used form 500hz and up, i understand that people like to take a driver which is a bit smaller then a fe167e; i don't really have to re-use this driver, but i like the high effency, and the smaller drivers of fostex don't have that high efficency which is a bit a pitty.

thanks everyone

[ultrachrome]

The crossover point is at 500Hz but due to the dipole hump the driver appears to be down only 5dB at 200Hz.

I would think cone tweaks would not affect dipole performance as the interaction between the dipole baffle and the driver occur well below 1kHz.