Well, to tell you an awful truth of mine: i don't give a sh.t about what musicians think about sound during session.
I'm the technician, i know where i want to go and musicians are most of the time more biased than anyone else...
I learned the hard way being bass player in metal bands in early 90's: guitarist and drummer thougths were no need for bass... until Metalica's Black Album was released and they sold million's copy BECAUSE there was at least some bass frequencies to dance to... lol.
Whatever... 4 WAYS. Next one for me will be 4 ways. Because it limits IMD. No needs for 120db spl for me but IMD limitation.
Depends of where you put xover freq but i think something like 20/80-150hz, 80-150/ 320-600, 320-600/ 1200-2400, 1200-2400 and up makes sense.
Keep high deplacement away from the 300/3k range where the 'most important' freq range is.
It offer a way to taylor some important parameter (to me) under control too: possibility to taylor membrane area to freq range to reproduce (minimum doubling of membrane area each time you drop an octave to limit distortion), control the way drivers/box interact for each freq band ( acoustical/mechanical), implement an acoustical design to have a constant and manageable directivity behaviour ( vertical/horizontal) over as much range as possible,...
All this sum up in the end. And dividing into multiple bands help. The issue being how it'll sum up as a global in the end.
I'm the technician, i know where i want to go and musicians are most of the time more biased than anyone else...
I learned the hard way being bass player in metal bands in early 90's: guitarist and drummer thougths were no need for bass... until Metalica's Black Album was released and they sold million's copy BECAUSE there was at least some bass frequencies to dance to... lol.
Whatever... 4 WAYS. Next one for me will be 4 ways. Because it limits IMD. No needs for 120db spl for me but IMD limitation.
Depends of where you put xover freq but i think something like 20/80-150hz, 80-150/ 320-600, 320-600/ 1200-2400, 1200-2400 and up makes sense.
Keep high deplacement away from the 300/3k range where the 'most important' freq range is.
It offer a way to taylor some important parameter (to me) under control too: possibility to taylor membrane area to freq range to reproduce (minimum doubling of membrane area each time you drop an octave to limit distortion), control the way drivers/box interact for each freq band ( acoustical/mechanical), implement an acoustical design to have a constant and manageable directivity behaviour ( vertical/horizontal) over as much range as possible,...
All this sum up in the end. And dividing into multiple bands help. The issue being how it'll sum up as a global in the end.
Magnepans are not dipole speakers, there polar response is all over the place, it's a flawed design. I have had them here in the same room I run dipole speakers and it simply cannot be compared to good dipole speakers. I did not like them at all, they need at least 3 meters behind them, any reflection from the back wall destroys the soundstage completely while I never had that problem with dipole speakers.
In my experience dipoles are easier to make behave than boxed drivers, it's the easiest way to get constant directivity and a consistent sound across the room.
Dipoles easier to have a constant directivity ( over a large freq range)?
It seems it's not something widely accepted upon pro designers, here is an eg from our missed SpeakerDave ( on second part of the interview but the whole is interesting read nevertheless):
https://www.tnt-audio.com/intervis/david_smith_e.html
Your description of Magnepans needing lot's of space to work as intended should be seen as the expected behavior from dipole source. I would question your dipole implementation from not needing this: don't take me wrong, i'm sure this sound good and your pleased with them but dipoles are dipoles... for not to have ER to blur everything you need space around them.
It seems it's not something widely accepted upon pro designers, here is an eg from our missed SpeakerDave ( on second part of the interview but the whole is interesting read nevertheless):
https://www.tnt-audio.com/intervis/david_smith_e.html
Your description of Magnepans needing lot's of space to work as intended should be seen as the expected behavior from dipole source. I would question your dipole implementation from not needing this: don't take me wrong, i'm sure this sound good and your pleased with them but dipoles are dipoles... for not to have ER to blur everything you need space around them.
Isn't that entirely subjective, imprecise, and borderline hearsay? I'm between shock and laughter.
Some suggestions.
Always active filters, and I understand that digital ones beat analog ones.
Separate cabinets. For the middle band I use two BLANDA MATT from ikea to make a sphere. I have custom 3D printed spere for Rall 70-10
Use the best speakers you can afford. This makes the difference. Currently I would use Purify 10", elepticor for mids and Raal 140-15 o better for highs.
Alternatives. Purify also works very well in mids. SB or Scan Speak berilium twitter
Always look at the speaker distortion.
Greetings and good luck
Always active filters, and I understand that digital ones beat analog ones.
Separate cabinets. For the middle band I use two BLANDA MATT from ikea to make a sphere. I have custom 3D printed spere for Rall 70-10
Use the best speakers you can afford. This makes the difference. Currently I would use Purify 10", elepticor for mids and Raal 140-15 o better for highs.
Alternatives. Purify also works very well in mids. SB or Scan Speak berilium twitter
Always look at the speaker distortion.
Greetings and good luck
Magnepans are not dipoles. Simply because the emit sound from both sides does not automatically make them dipoles.
I have measured them and they do not measure like dipoles at all, it's a flawed design that goes back to the early 80's when of-axis behaviour let alone constant directivity was not something loudspeaker designers cared about. A proper designed dipole on the other hand has very good of-axis behaviour and is the easiest way to get fullrange constant directivity down to the lower midrange/upper bass, something otherwise only possible with very big horns/waveguides or proper designed cardioid constructions
No. Read my posts and try again.
I am saying newbies who want a low risk, fast track to a proven design for the lowest cost without having to spend the time to become a "proficient loudspeaker designer" are best served by cloning a proven design from an expert group of audio engineers.
So I hope you understand better now, also I do hope that expressing is ok with you? If I dont have your blessing how can I sleep at night?
Hope I dont come across as too weird or, god forbid defensive... So many sensitive souls on this forum... Who knew?
Some good stuff in the Bruno Putzeys & Eelco Grimm paper you linked to, thanks again for posting it :https://www.grimmaudio.com/wp-content/uploads/speakers.pdf
"Loudspeaker design is an “ill-posed problem” because some requirements are in conflict while others can’t even be definitively stated. Working such problems into an acceptable compromise is done not by starting from what we want to get done, but from the things we want to avoid and prioritizing those."
That is brilliant!
Re the OP and jheoaustin's design, I think this section is particularly relevant, although it is not what you want to hear!
" It looks perfectly normal by audiophile standards: a pair of 8” woofers in a 9” wide cabinet with a ribbon tweeter between them. Vertically, the wide spacing of the woofer results in strong lobing. The ribbon might have good horizontal dispersion (never omitted in data sheets), it is very directive vertically (never published). The narrow cabinet places the baffle step frequency in the middle of the vocal range, resulting in a dry top end and woolly low-mids. Conversely, doing it right means: a single unit per “way”, spaced closely together on a baffle of at least 20” wide. Now, wide baffles have a reputation of less than stellar imaging. This is because diffraction artefacts become audible as distinct reflections whereas on narrow baffles they are perceived only as colouration. The solution is simple enough: generously round off the edges."
" It looks perfectly normal by audiophile standards: a pair of 8” woofers in a 9” wide cabinet with a ribbon tweeter between them. Vertically, the wide spacing of the woofer results in strong lobing. The ribbon might have good horizontal dispersion (never omitted in data sheets), it is very directive vertically (never published). The narrow cabinet places the baffle step frequency in the middle of the vocal range, resulting in a dry top end and woolly low-mids. Conversely, doing it right means: a single unit per “way”, spaced closely together on a baffle of at least 20” wide. Now, wide baffles have a reputation of less than stellar imaging. This is because diffraction artefacts become audible as distinct reflections whereas on narrow baffles they are perceived only as colouration. The solution is simple enough: generously round off the edges."
A simple question. If you don't know basics how do you trouble shoot if you have an issue? Many of us started with cones and moved on as we gained more knowledge. In retrospect when I was "cloning" I wish I knew then what I know now. Would have made things a lot easier. It's not a 2 step process to do it right takes work.
All of that maters. If you don't look at that as part of your design you are not going to get the best out of the driver set. And if you are cloning better hope your clone designer did.
Polar response and cabinet edge diffraction can't be fixed with any EQ be it analog or digital so if there are issues you have to change the original design to effect them.
Rob
(7) Size matters... Try to use the largest Sd you can to cover any given frequency band, this can be one large driver or multiple smaller drivers.
I see a conflict
Easy mistake to make pico, confusing Eq with driver Sd... DSP and driver cones are so similar...
The clue is in reading my post ie "try to use the largest Sd you can to cover any given frequency band, this can be one large driver or multiple smaller drivers." Note the words "Sd and drivers".
Now the complicated part "You can Eq the baffle edge issues away, but it will be easier the wider the cabinet, less Eq required and less is always better."
Note the words " Eq, and less Eq required is always better"
Good luck.
The clue is in reading my post ie "try to use the largest Sd you can to cover any given frequency band, this can be one large driver or multiple smaller drivers." Note the words "Sd and drivers".
Now the complicated part "You can Eq the baffle edge issues away, but it will be easier the wider the cabinet, less Eq required and less is always better."
Note the words " Eq, and less Eq required is always better"
Good luck.
Thanks @jheoaustin !
In my 3-Ways 375L, the transistion frequencies are 400Hz and 5000Hz. This combination works surprisingly very fine with this crossover design I ended with :
But I want to experiment with an intermediary loudspeaker circa 5000Hz, because the tweeter of the 375L is recommended to be cut at 5000Hz, and also because I would like to narrow the bandwidth of each speaker to improve further the transistion characteristics, having the speakers cuts in a more linear area of FR and dispersion.
In my 3-way 375L, the tweeter is a Beyma CP21/F.
T
I didn't stress the 'always' enough.
So Sd always better/ less EQ always better
..er, no, best!!
No, you didn't write it.... but: closely...how closely? We have that 20" number...there are refractions/ diffractions... wavelengths from 2cm to 17m.
Put that together with 'as much as Sd possible ' and you end up with...
Yes, 3 way
And three is the...
Magic
eh, nope, thats beyond EQ. To fix cabinet (edge) issues you must fix the cabinet. This is where you need to geek out the structure, and measure polars, and the stuff you dismissed earlier when hifijim tried to bring it up.
EQ will affect frequency response to all directions, while size and shape of the structure makes response to vary to all directions. Bafflestep one can compensate, but the same diffraction effect on shorter wavelengths makes response to vary to all directions., and the wider bandwidth the bigger the baffle, unless it approaches size of your room. You can "fix" one axis with EQ, but others get worse. If you do this, for one axis, you get locked into one toe-in which limits positioning freedom, sound changes if you move a little, so it's not a fix other than on computer screen with one axis measurement. EQ affects power response so overall sound changes due to that as well.
One could be better not using EQ here, to fix edge diffraction related ripple. It is an issue that doesn't seem to bother many, like the dudes in the video series, but stuff like this, acoustics, is something one might want to look into at some point, sooner or later, if you pursue good sound it's one thing one can "fix". This stuff is in the Putzney quote as well. After all, it's the acoustics we listen to, and our hearing system processes for to a conscious perception of it
edit. this stuff depends a lot on context, wheter something is more or less important. For example edge diffraction might not be that bad if room early reflections overwhelm the direct sound, so depends on positioning and listening distance. These then might depend on many things, like practical issues, whats the recording, whether you need to concentrate on the music or do studying and all kinds of stuff and listen further out with the earlyreflections. One might go close enough, fpr proper engagement, but back out because sound wasn't very good. Here, it light have been partly by diffraction, partly by any issue like DI or amplifier noise, or just the recording that was mixed to be listened with strong early reflections. So, it"s something one might consider more or less important. It's good to know this stuff though, so one can design a system that tailor fits to ones context.
Last edited:
Alex, is this the quote/comment for me? I hope not. I just want to point out(not to you, but to the one who wrote that) that I never said I'd use any ribbons in this, and never said I'd have a wide spacing between 2 woofers, or the woofers and the midbass. Opposite to the quote, I think I said I heard only 2 ribbons I liked sonically, and all others I heard including my MG 1.7i weren't so, and that I wanted to minimize the baffle area, which will indirectly results in smaller spacings between drivers. Sorry, I am just too lazy to go back to dozens of posts to find who wrote the quoted, and reply direct to him. You know I like your philosophy and agree on at least some parts of it.
This is where wide dispersion speakers and room treatment come into play. The reverberation field within the room needs to be as spectrally similar (as possible) to the direct sound field, so that the on-axis EQ would work (better) when off-axis ...
Thanks @jheoaustin !
In my 3-Ways 375L, the transistion frequencies are 400Hz and 5000Hz. This combination works surprisingly very fine with this crossover design I ended with :
But I want to experiment with an intermediary loudspeaker circa 5000Hz, because the tweeter of the 375L is recommended to be cut at 5000Hz, and also because I would like to narrow the bandwidth of each speaker to improve further the transistion characteristics, having the speakers cuts in a more linear area of FR and dispersion.
In my 3-way 375L, the tweeter is a Beyma CP21/F.
T
Great. I happened to talk to someone knowledgeable offline to hear that he'd crossover Satori around 4~5kHz on the top. I didn't get a clear answer about the lower crossover, though, and I am still hoping for 500Hz or around...