Th
I posted earlier studies of different shapes.
However l feel it is appropriate that you sketch out on paper your idea of how you think this is going to look and post it for general discussion.
A few thoughts.
Let’s assume that for effective diffraction control you decide on using two cylinders placed in a vertical plane.
The end of each cylinder is terminated with a round over making the diameter 120 mm.
So let’s stop and look at this.
To allow the round overs to function you will need to allow an air gap between each of the cylinders. Let’s assume that’s 50mm.
The cylinders are relatively short at 100 mm and are fastened at the rear to a vertical timber frame.
The overall dimensions of the loudspeaker are 200mm wide and some 120 cm tall.
Looking at this the diffraction control is only necessary in the horizontal plane. The amplitude ripples are more likely with a baffle of equal dimensions unless explicitly designed with FEA to form an oblate spherical shape. That however is outside the scope of a diy loudspeaker project for the purposes of this discussion.
We know that spreading out the mid and hf drivers vertical attracts of problems with the acoustics. The minimum listening distance gets pushed back and there are interference patterns in the vertical plane. Off axis differences in the driver vertical plane emerge with shifts in listener height.
This is all being considered because you wish to construct a four way loudspeaker.
The vertical placement of the four drivers leaves open several possibilities with some benefit. You need to work through that before you start looking too closely at baffles abd baffle shapes.
To break all this down you really need to work though your priorities and determine which are going to work together and which are conflicting.
From there I would suggest you determine which part of the project your going to tackle and start on that. From there your options will naturally narrow down to what is feasible.
As far as the above topic goes you need to be prepared to buy some drivers and start some mock ups of how it’s going to come together. It might take a dozen mockup’s before you settle on the one that delivers what youf feel comfortable with.
I can certainly assure you with testing and l may acquire some of your proposed drivers to investigate them more closely.
I think it would be a good idea if you describe the success and limitations of your listening space. That is probably the key to determining what’s appropriate.
I don’t know.
I posted earlier studies of different shapes.
However l feel it is appropriate that you sketch out on paper your idea of how you think this is going to look and post it for general discussion.
A few thoughts.
Let’s assume that for effective diffraction control you decide on using two cylinders placed in a vertical plane.
The end of each cylinder is terminated with a round over making the diameter 120 mm.
So let’s stop and look at this.
To allow the round overs to function you will need to allow an air gap between each of the cylinders. Let’s assume that’s 50mm.
The cylinders are relatively short at 100 mm and are fastened at the rear to a vertical timber frame.
The overall dimensions of the loudspeaker are 200mm wide and some 120 cm tall.
Looking at this the diffraction control is only necessary in the horizontal plane. The amplitude ripples are more likely with a baffle of equal dimensions unless explicitly designed with FEA to form an oblate spherical shape. That however is outside the scope of a diy loudspeaker project for the purposes of this discussion.
We know that spreading out the mid and hf drivers vertical attracts of problems with the acoustics. The minimum listening distance gets pushed back and there are interference patterns in the vertical plane. Off axis differences in the driver vertical plane emerge with shifts in listener height.
This is all being considered because you wish to construct a four way loudspeaker.
The vertical placement of the four drivers leaves open several possibilities with some benefit. You need to work through that before you start looking too closely at baffles abd baffle shapes.
To break all this down you really need to work though your priorities and determine which are going to work together and which are conflicting.
From there I would suggest you determine which part of the project your going to tackle and start on that. From there your options will naturally narrow down to what is feasible.
As far as the above topic goes you need to be prepared to buy some drivers and start some mock ups of how it’s going to come together. It might take a dozen mockup’s before you settle on the one that delivers what youf feel comfortable with.
I can certainly assure you with testing and l may acquire some of your proposed drivers to investigate them more closely.
I think it would be a good idea if you describe the success and limitations of your listening space. That is probably the key to determining what’s appropriate.
Yes an older passive design with several versions over many years. The ratio of cabinet weight vs mms is not good? Why?
Rob
I heard an L250 yesterday afternoon. I wax pleasantly surprised by its capabilities.
Manufacturers test data
https://sbacoustics.com/wp-content/uploads/2022/03/2.5in-SATORI-MD60N-6.pdf
Looking at the above data there is no mention of the test conditions. But it looks good.
However, in an impartial review below in Voice Coil with full documentation and disclosure we have these graphs.
https://audioxpress.com/article/tes...me-midrange-from-the-sb-acoustics-satori-line
Looking at the test baffle dimensions of 15 x 6 inches l was able to deduce that the 1/2 wavelength from the centre of the baffle is . This is causing this trough. It’s irrefutable. But the off axis 30 degree plot is smooth. That is a problem that cannot be resolved with EQ without messing up the important room response.
Why is the 30 degrees off axis smoother? My guess is the geometry of the dome.
Looking at the normalised graphs it’s cleaned up but what emerges is a less than ideal situation when attempting to obtain off axis continuity with the tweeter in the crossover region. There is 2 db dip in the 30 degree off axis response at 1,500 hertz only to see it peak back to 00.00 db at 2800 hertz.
If you live and die by look at Erin’s Klippel Scanner measurements which people are these days you might look at a simpler system.
My thoughts are that you are going to have to try several approaches with the surrounding baffle on the Satori MD60 before coming up with a satisfactory compromise. It’s a nice driver but it won’t win without some real work.
I would also be looking into a plan B where someone has incorporated this driver such as the design below. Email Toels and pick his brain on this driver.
Below is a build with the mid dome outline the challenge and the solutions. Also attached are some alternatives.
http://www.troelsgravesen.dk/SBA-7MD.htm
http://www.troelsgravesen.dk/SBA-741.htm
http://www.troelsgravesen.dk/SBA-16-MTM.htm
http://www.troelsgravesen.dk/SBAcoustics-10.htm
https://sbacoustics.com/wp-content/uploads/2022/03/2.5in-SATORI-MD60N-6.pdf
Looking at the above data there is no mention of the test conditions. But it looks good.
However, in an impartial review below in Voice Coil with full documentation and disclosure we have these graphs.
https://audioxpress.com/article/tes...me-midrange-from-the-sb-acoustics-satori-line
Looking at the test baffle dimensions of 15 x 6 inches l was able to deduce that the 1/2 wavelength from the centre of the baffle is . This is causing this trough. It’s irrefutable. But the off axis 30 degree plot is smooth. That is a problem that cannot be resolved with EQ without messing up the important room response.
Why is the 30 degrees off axis smoother? My guess is the geometry of the dome.
Looking at the normalised graphs it’s cleaned up but what emerges is a less than ideal situation when attempting to obtain off axis continuity with the tweeter in the crossover region. There is 2 db dip in the 30 degree off axis response at 1,500 hertz only to see it peak back to 00.00 db at 2800 hertz.
If you live and die by look at Erin’s Klippel Scanner measurements which people are these days you might look at a simpler system.
My thoughts are that you are going to have to try several approaches with the surrounding baffle on the Satori MD60 before coming up with a satisfactory compromise. It’s a nice driver but it won’t win without some real work.
I would also be looking into a plan B where someone has incorporated this driver such as the design below. Email Toels and pick his brain on this driver.
Below is a build with the mid dome outline the challenge and the solutions. Also attached are some alternatives.
http://www.troelsgravesen.dk/SBA-7MD.htm
http://www.troelsgravesen.dk/SBA-741.htm
http://www.troelsgravesen.dk/SBA-16-MTM.htm
http://www.troelsgravesen.dk/SBAcoustics-10.htm
Attachments
I did a few basic sims with the excel diffraction model.
The best l could obtain is the attached.
The modeller does not do a circle.
It will do your head in.
Placing a 1 inch tweeter was no problem at all. The longer baffle offers a smoother overall response and better bottom end.
Anything with equal dimensions was not favourable.
The radius edge helps a bit but it’s not a cure .
The best l could obtain is the attached.
The modeller does not do a circle.
It will do your head in.
Placing a 1 inch tweeter was no problem at all. The longer baffle offers a smoother overall response and better bottom end.
Anything with equal dimensions was not favourable.
The radius edge helps a bit but it’s not a cure .
Attachments
https://www.linkwitzlab.com/diffraction.htm
The above link is a very good explanation of circular baffles and other variables effecting diffraction and baffle step
The above link is a very good explanation of circular baffles and other variables effecting diffraction and baffle step
No but can be cloned or purchased used. 4 different versions over 15-20 years.
Rob
https://www.audioheritage.org/html/profiles/jbl/l250.htm
I made in interesting observation.
In the same room are Duntech Sovereigns which are massive. They have dual 12 inch woofers and a full MTM array. But they don’t sound as dynamic and punchy as the L250.
This is because while the Sovereigns go lower they are overall a less efficient design using a sealed enclosures. The L250 has a rated sensitivity of 94db. In comparison the Sovereign is rated at 90 db 1 watt 1 metre (8 ohms according to the attached document
http://duntech.com/products/sovereign.html
https://support.jbl.com/on/demandwa...n/default/dw04d425fe/pdfs/L250^JBL^om^^^^.pdf
The L250 has a faster more punchy bass reverberant sound.
The LE14 woofers and the 8 inch mid driver really capture the bass lines. The L250 drivers are designed to capture transient peaks.
The mid bass is crisp and fast using a large Krell power amp. The Sovereign does have a deep sound stage but it misses the impact and resolution of percussion instruments.
Some may disagree but the proof is in the listening. The owner has had a number pf L250’s. He made the remark and l agree having heard the Sovereigns.
Troel’s made similar observations about his loudspeaker One which was inspired by the JBL 4343.
This brings me to the conclusion that as individuals we don’t all agree on how we like to listen to our music.
I prefer to be close to the front of a live performance. I like to feel the impact and the intensity of a band. To be immersed in the raw energy. I feel the same emotional connection sitting not far from an acoustic guitarist.
I prefer loudspeakers that can translate this scenario. I believe some JBL system fulfil that criteria.
I am curious how the TO likes to listen to his music?
We are all different and that is okay.
In the same room are Duntech Sovereigns which are massive. They have dual 12 inch woofers and a full MTM array. But they don’t sound as dynamic and punchy as the L250.
This is because while the Sovereigns go lower they are overall a less efficient design using a sealed enclosures. The L250 has a rated sensitivity of 94db. In comparison the Sovereign is rated at 90 db 1 watt 1 metre (8 ohms according to the attached document
http://duntech.com/products/sovereign.html
https://support.jbl.com/on/demandwa...n/default/dw04d425fe/pdfs/L250^JBL^om^^^^.pdf
The L250 has a faster more punchy bass reverberant sound.
The LE14 woofers and the 8 inch mid driver really capture the bass lines. The L250 drivers are designed to capture transient peaks.
The mid bass is crisp and fast using a large Krell power amp. The Sovereign does have a deep sound stage but it misses the impact and resolution of percussion instruments.
Some may disagree but the proof is in the listening. The owner has had a number pf L250’s. He made the remark and l agree having heard the Sovereigns.
Troel’s made similar observations about his loudspeaker One which was inspired by the JBL 4343.
This brings me to the conclusion that as individuals we don’t all agree on how we like to listen to our music.
I prefer to be close to the front of a live performance. I like to feel the impact and the intensity of a band. To be immersed in the raw energy. I feel the same emotional connection sitting not far from an acoustic guitarist.
I prefer loudspeakers that can translate this scenario. I believe some JBL system fulfil that criteria.
I am curious how the TO likes to listen to his music?
We are all different and that is okay.
I think one problem with the Duntech Sovereigns is the excessive space of the 17W 75s which causes timing issues for the listener. I played around with a few WMTMM designs and was just never satisfied with the sound.
I suspect I would choose the L250 over the Sovereign if given the choice.
I suspect I would choose the L250 over the Sovereign if given the choice.
I've been and what looking at what the best commercial systems look like based the initial idea of deal 8 inch woofers.
Using the Wilson Audio Sasha V I have negatively engineered the ball park bass enclosure volume and done some baffle step simulations. I then did a simulation with a 6.5 inch driver and an 8 inch wide baffle.
The results.
My estimate is the bass enclosure is some 30 inches high, 14.5 inches wide and about 24 inches deep
Allowing for 1 inch mdf panels these dimensions provide an overall volume of about 120L or 4.5 cu ft 3.
Thinking about it and looking at the form factor the width and height seem proportional whereas the 8 inch width doesn't seem proportional. I would therefore suggest that if 8 inch woofers are desired then the 14.5 inch width might in fact be a blessing in disguise. The engineering upside is that this enclosure allow full exploration of several different enclosure tunings to maximise the potential of any driver.
Next I looked at the baffle step when synthesised to incorporate the combined room boundary and room gain. By placing the 8 inch drivers in either the top and bottom location it reveals that Wilson have a well thought out acoustic solution to deliver a smooth in room response with both drivers operating over different bandwidth's.
Conclusions
I feel the above ball park negative engineering of the Wilson Audio Sasha V acoustic solution proves the likely success of a dual 8 inch system using the above overall bass enclosure dimensions. A narrower enclosure of 8 inches would reduce the internal dimensions to 6 inches and have significant impact of the usable volume of the enclosure by about 52%.
Of course the TO may prefer to use the cylinders. But this will greatly reduce the available enclosure volume. Looking at the Linkwitz technical guide to diffraction a circular baffle appears to offer little if any advantage over a well proportioned rectangular or tapered rectangular baffle.
Using the Wilson Audio Sasha V I have negatively engineered the ball park bass enclosure volume and done some baffle step simulations. I then did a simulation with a 6.5 inch driver and an 8 inch wide baffle.
The results.
My estimate is the bass enclosure is some 30 inches high, 14.5 inches wide and about 24 inches deep
Allowing for 1 inch mdf panels these dimensions provide an overall volume of about 120L or 4.5 cu ft 3.
Thinking about it and looking at the form factor the width and height seem proportional whereas the 8 inch width doesn't seem proportional. I would therefore suggest that if 8 inch woofers are desired then the 14.5 inch width might in fact be a blessing in disguise. The engineering upside is that this enclosure allow full exploration of several different enclosure tunings to maximise the potential of any driver.
Next I looked at the baffle step when synthesised to incorporate the combined room boundary and room gain. By placing the 8 inch drivers in either the top and bottom location it reveals that Wilson have a well thought out acoustic solution to deliver a smooth in room response with both drivers operating over different bandwidth's.
Conclusions
I feel the above ball park negative engineering of the Wilson Audio Sasha V acoustic solution proves the likely success of a dual 8 inch system using the above overall bass enclosure dimensions. A narrower enclosure of 8 inches would reduce the internal dimensions to 6 inches and have significant impact of the usable volume of the enclosure by about 52%.
Of course the TO may prefer to use the cylinders. But this will greatly reduce the available enclosure volume. Looking at the Linkwitz technical guide to diffraction a circular baffle appears to offer little if any advantage over a well proportioned rectangular or tapered rectangular baffle.
Attachments
Last edited:
I suspect that the wall thickness is substantially more than 1 inch. Wilson speakers are known for the elaborate cabinet construction, to the point of being over the top. The total weight of the speaker is 245 lb. The total weight of the 4 drivers is probably less than 35 lb. Where is all the extra weight coming from?
Try using 50 mm in your reverse engineering. The reality might be even thicker. It is also likely that MDF is not the cabinet material, but something more dense.
I am afraid I am too short in these acoustics knowledge to understand your points correctly, but it appears that there could be a fundamental problem in this subsystem of tweeter/midhigh/midbass approach that is unrelated to the baffle design. Am I understanding it correctly?
You are right that Satori dome mid is cheap enough to do some experiments. While I am looking at bass module options, I think I can start from the tweeter module to build and place it on top of my Wilmslow as a modular test, and then move on ro midbass to try it out somehow. Midbass tryout requires some changes in my electronics/DSP, so it will have to wait a bit. It's good that I can buy the drivers and work on different prototypes, reusing one or two drivers.
@jheoaustin - There is not universal agreement on the issue of driver spacing. Some people feel that closer is better. Others (like me) believe the situation is more complicated.
First, my following comments apply to crossovers above 300 Hz. Below that point, room effects tend to dominate.
If the center to center distance (CTC) between two drivers can be less than 1/4 wavelength of the crossover frequency, this is the best case. The two drivers will act as a single source. We can relax the requirement to 1/3 wavelength and still achieve a nearly single point source.
However, if a CTC of 1/3 wavelength is not possible due to the size of the drivers, then closer is not necessarily better. This is almost always the case with midrange-tweeter crossovers. There will always be vertical cancelations if the CTC is above 1/3 WL, and so the choice we face is not how to avoid vertical cancellations, but how to manage the shape and frequency of the cancellations Empirical evidence and simulations show that 1.2 x wavelength, combined with minimal baffle size and 2nd or 4th order crossover slopes, is a good compromise.
https://www.diyaudio.com/community/...ivers-a-new-3-way-project.413182/post-7708971
j.
I also prioritise realistic dynamics which IMO are far more important than dispersion and diffraction related issues.
Two of the keys to life like dynamics are high sensitivity and high quality pro drivers with low Mms / high Bl and cotton/fabric surrounds.
The Duntech uses what look like "audiophile" (contradiction in terms
) heavy coned / rubber surround drivers... Never a good idea.
Yes unrelated.. the problem is the vertical length of the driver line up in achieving a 400-4000 hertz range with the tweeter to 20,000.
Presumably you use an 8 inch woofer, the 60mm mid done with a 5 inch top plate and a tweet with a 4 inch top plate.
Woofer 8 inch
Space. 6 inch
Mid dome 5 inches
Space 3 inches
Tweeter 4 inches
That’s at least 17 inch and probably 26 inches in practice allowed from spacing.
Is it best practice to have the human voice (mid range) spread a 26 inch length?
This will pose compromises. Notably the vertical polar pattern. Secondly obtaining even off axis curves.
If l can indulge you that in the quest for perfect drivers you might be fracturing the sound stage and imaging.
Taking a look at best practice and maybe a slight compromise in the midrange a 6 inch midrange driver and a narrow face plate tweeter might more easily put a smile on your face.
Liets set the crossover points at 400 and 3000 hertz.
Dual 8 inch woofer
Space 4 inches
Mid range 6.5 iinch
space 2 inches
Tweeter 3 inches
In this configuration your mid range and top end are within 11.5 inches
The eyes often do the hearing too.
what l am referring to is the practical realisation of your proposal versus what successful design looks like .
Examples are the Wilson Sasha V crossover to 7 inch mid range 400 hetz
The B&W 801 D4 crossover 400 hertz
This approach simplifies the woofer configuration.
In both cases the whole midrange band is elevated Weill above the floor and defined in a contained space.
I appreciate chasing the perfect drivers set is important but in the end what are your ears going to say about it?
Attachments
Correction: The Wilson Audio systems crossover the 7 inch mid cone between 150-250 hertz and up to 2800 hertz.
The fabrication of the egg shaped mid enclosure poses a diy challenge.
One idea is to use a 10 inch cardboard tube and carefully cut in at the point where the diameter of the cut is 8 inches.
Using an 8 inch mdf disk mitre the edge to a round over to match the tube diameter at the cut. This may require laminated layers of mdf or plywood to build out the curves surface to match the midrange cone (6.5 - 7.00 inches)
With the driver cutout made glue the baffle in with liquid nail.
Add your favourite goop to damp the tube. Then repeat the process on the rear.
Another idea is to use a sports inflatable ball to make a mould for shapes that can be glued together.
See the attached sketch
The red lines are dampening layers … couldn’t resist the urge. Red
The fabrication of the egg shaped mid enclosure poses a diy challenge.
One idea is to use a 10 inch cardboard tube and carefully cut in at the point where the diameter of the cut is 8 inches.
Using an 8 inch mdf disk mitre the edge to a round over to match the tube diameter at the cut. This may require laminated layers of mdf or plywood to build out the curves surface to match the midrange cone (6.5 - 7.00 inches)
With the driver cutout made glue the baffle in with liquid nail.
Add your favourite goop to damp the tube. Then repeat the process on the rear.
Another idea is to use a sports inflatable ball to make a mould for shapes that can be glued together.
See the attached sketch
The red lines are dampening layers … couldn’t resist the urge. Red
Attachments
Last edited:
Thank you for your clear points and explanation here. One that really hit me hard was 'Is it best practice to have the human voice (mid range) spread a 26 inch length?'. I tried hard to think of some ways to mitigate the issue you pointed out, and here are 3 I came up with:
- Use D2004/6020 tweeter whose motor assembly diameter is 55mm. I ditched it earlier due to its dip around 15kHz(I think), but I guess that's something DSP can correct. Will reducing one distance from 80mm to 55mm help appreciably?
- Raise the crossover frequency from 4000Hz to 4500 ~ 5500Hz, so there is less voice component produced by the tweeter. Satori can support that band on axis, but 30-degree response seems to struggle a bit past 5000Hz. How much do you think it can be of help?
- Use the 5" illuminator midbass to reduce another CTC distance. This may call for 4x 7" instead of 2x 8"....
Human voice fundamentals are 200-2000Hz, but harmonics and noises spread to 4000hz, with much lower spl.
Many are happy cross at 400 and 2500Hz. LR2 gives smooth 180 deg phase change but LR4 turns 360deg and in narrower band.
My modded Avalanche active uses LR2 and 4"mid - only moderate eq was needed with Audax HM100Z0. CTC is 12cm (5")
Many are happy cross at 400 and 2500Hz. LR2 gives smooth 180 deg phase change but LR4 turns 360deg and in narrower band.
My modded Avalanche active uses LR2 and 4"mid - only moderate eq was needed with Audax HM100Z0. CTC is 12cm (5")