for those new to this train wreck V2 was here:
and now we are back to 5 way ...
actually i forgot to draw enclosures around 10" woofers ( red ) so just imagine it is there ...
this design retains the 4 boxes stacked on top of each other from V2 each with a 18" sub in it - because that just made sense ...
and the reason i had to redesign V2 is because it was based around 10" Radian Planar and it turned out that it doesn't use a Kapton diaphragm but some mother material that is likely less heat resistant as evidenced by fairly low power rating for such a large driver ...
so now we're back to the 8" Radian Planar which DOES use kapton and thus has higher power density per square inch of diaphragm ( about double that of the 10" radian ) because it can take higher temps ... this driver is shown as CYAN TRIANGLE in the sketch ... of course only the flat front of the triangle is actually the driver - the rear of the triangle is a chamber for it that is triangular so the output of MAGENTA midrange ( 5" cone ) can go around it ...
complete driver complement :
4.5 + khz - Aurum Cantus G3
https://www.parts-express.com/Aurum-Cantus-G3-Ribbon-Tweeter-276-412?quantity=1
1.2 khz - 4.5 khz - Radian 8”
https://radianaudio.com/collections/ribbon/products/lm8k-wide-band-planar-ribbon-transducer
320 hz - 1.2 khz - 5” B&C Bucket mid ( available from multiple stores )
https://www.bcspeakers.com/en/products/lf-driver/5/8/5NSM38
80 hz - 320 hz - 10” Beyma 10MC700ND ( sold by usspeaker )
https://www.beyma.com/speakers/Fich...rs-data-sheet-low-mid-frequency-10MC700Nd.pdf
16 hz - 80 hz - 18” NTLW5000 ( sold on Amazon )
https://www.eighteensound.it/en/products/lf-driver/18-0/8/18ntlw5000
messed up the title - it was supposed to say " please critique this " like the original thread.
it became obvious to me that i can completely eliminate one of 5 frequency bands from original design without much detriment so i did. here is the new design ( this is a horizontal cross section / view of array from the top ). the top of image is a waveguide that is forward facing towards the listener. the bottom of the image is a rear-firing subwoofer.
for reference here is the original thread when the design was 5 way, which is embarrassing now because of how much more...
it became obvious to me that i can completely eliminate one of 5 frequency bands from original design without much detriment so i did. here is the new design ( this is a horizontal cross section / view of array from the top ). the top of image is a waveguide that is forward facing towards the listener. the bottom of the image is a rear-firing subwoofer.
for reference here is the original thread when the design was 5 way, which is embarrassing now because of how much more...
and now we are back to 5 way ...
actually i forgot to draw enclosures around 10" woofers ( red ) so just imagine it is there ...
this design retains the 4 boxes stacked on top of each other from V2 each with a 18" sub in it - because that just made sense ...
and the reason i had to redesign V2 is because it was based around 10" Radian Planar and it turned out that it doesn't use a Kapton diaphragm but some mother material that is likely less heat resistant as evidenced by fairly low power rating for such a large driver ...
so now we're back to the 8" Radian Planar which DOES use kapton and thus has higher power density per square inch of diaphragm ( about double that of the 10" radian ) because it can take higher temps ... this driver is shown as CYAN TRIANGLE in the sketch ... of course only the flat front of the triangle is actually the driver - the rear of the triangle is a chamber for it that is triangular so the output of MAGENTA midrange ( 5" cone ) can go around it ...
complete driver complement :
4.5 + khz - Aurum Cantus G3
https://www.parts-express.com/Aurum-Cantus-G3-Ribbon-Tweeter-276-412?quantity=1
1.2 khz - 4.5 khz - Radian 8”
https://radianaudio.com/collections/ribbon/products/lm8k-wide-band-planar-ribbon-transducer
320 hz - 1.2 khz - 5” B&C Bucket mid ( available from multiple stores )
https://www.bcspeakers.com/en/products/lf-driver/5/8/5NSM38
80 hz - 320 hz - 10” Beyma 10MC700ND ( sold by usspeaker )
https://www.beyma.com/speakers/Fich...rs-data-sheet-low-mid-frequency-10MC700Nd.pdf
16 hz - 80 hz - 18” NTLW5000 ( sold on Amazon )
https://www.eighteensound.it/en/products/lf-driver/18-0/8/18ntlw5000
so it's 5-way and each frequency band is about 2 octaves for a total of 10 octaves of audible range
furthermore every frequency band uses a different type of transducer
true ribbon, planar, and three frequency bands are covered by neodymium cones ... wait, how is that different ?
well it's different because the 5" uses an outside ( of the voice coil ) ring magnet motor in a sealed bucket design which is optimal for midrange
the 10" uses an inside slug magnet motor that is optimal for woofers
and the 18" uses a push pull motor optimal for subwoofers
pretty proud of this. i had to somehow justify having 5 frequency bands and i did it.
furthermore every frequency band uses a different type of transducer
true ribbon, planar, and three frequency bands are covered by neodymium cones ... wait, how is that different ?
well it's different because the 5" uses an outside ( of the voice coil ) ring magnet motor in a sealed bucket design which is optimal for midrange
the 10" uses an inside slug magnet motor that is optimal for woofers
and the 18" uses a push pull motor optimal for subwoofers
pretty proud of this. i had to somehow justify having 5 frequency bands and i did it.
so V3 iteration is all about power handling ...
the VHF ( very high frequency ) had absurd power already in V2 and so is unchanged. Aurum Cantus G3 are 100 db efficient and with a pure aluminum diaphragm there is no plastic to melt so they can take power as well and the heat is spread across the entire ribbon surface which is exposed to outside air for cooling. essentially power handling is a non-issue here. instead if the ribbon breaks it will be from the subwoofers or woofers blowing it out with sound pressure.
subwoofer section is also unchanged from V2 and is only different from V1 in physical layout but not in component choice or tuning.
the impetus for redesigning V2 as i said was to replace Radian 10" with Radian 8" because it has twice the power density with Kapton film rather than Polyester ... but the 8" radian can't go down to 500 hz as the 10" did in V2 and in fact although it has 2X power density it doesn't have any more power - it is simply half the width with same power - so to actually increase overall power handling of the system we have to raise the crossover point from 500 hz to 1.2 khz ...
because the planar no longer has to reproduce 500 hz to 1.2 khz it has less work to do but still has the same power handling thus overall speaker capacity goes up ...
for 320 hz to 1.2 khz i chose B&C 5" bucket because it's the smallest prosound bucket midrange. a bucket midrange has its enclosure formed by the cast aluminum basket which means heat from the motor is transferred directly to outside air rather than being trapped inside a plywood box. aluminum conducts heat while plywood is an insulator. this 5" also has 1.5" VC which is very large for such a small driver as well it is also quite efficient though not as efficient as B&C claims.
for midbass i chose 10" beyma because it has a good mix of Xmax, Efficiency, Power Handling and FS for that frequency range and is still small enough for packaging and directivity purposes. i have also used Beyma several times in the past and was always happy with their drivers. it is also one of their newer and more fully featured / high performance drivers.
the VHF ( very high frequency ) had absurd power already in V2 and so is unchanged. Aurum Cantus G3 are 100 db efficient and with a pure aluminum diaphragm there is no plastic to melt so they can take power as well and the heat is spread across the entire ribbon surface which is exposed to outside air for cooling. essentially power handling is a non-issue here. instead if the ribbon breaks it will be from the subwoofers or woofers blowing it out with sound pressure.
subwoofer section is also unchanged from V2 and is only different from V1 in physical layout but not in component choice or tuning.
the impetus for redesigning V2 as i said was to replace Radian 10" with Radian 8" because it has twice the power density with Kapton film rather than Polyester ... but the 8" radian can't go down to 500 hz as the 10" did in V2 and in fact although it has 2X power density it doesn't have any more power - it is simply half the width with same power - so to actually increase overall power handling of the system we have to raise the crossover point from 500 hz to 1.2 khz ...
because the planar no longer has to reproduce 500 hz to 1.2 khz it has less work to do but still has the same power handling thus overall speaker capacity goes up ...
for 320 hz to 1.2 khz i chose B&C 5" bucket because it's the smallest prosound bucket midrange. a bucket midrange has its enclosure formed by the cast aluminum basket which means heat from the motor is transferred directly to outside air rather than being trapped inside a plywood box. aluminum conducts heat while plywood is an insulator. this 5" also has 1.5" VC which is very large for such a small driver as well it is also quite efficient though not as efficient as B&C claims.
for midbass i chose 10" beyma because it has a good mix of Xmax, Efficiency, Power Handling and FS for that frequency range and is still small enough for packaging and directivity purposes. i have also used Beyma several times in the past and was always happy with their drivers. it is also one of their newer and more fully featured / high performance drivers.
for array purposes i always start out my search with the smallest drivers possible - as small as 2" but inevitably i find that they do not have the build quality or performance i am looking for ...
i then i have to go up 1 inch at a time and examine the options at every size level ...
generally the larger the driver the more advanced features you are able to have in it such as cast aluminum baskets, inside-outside voice coils, vented VC gap and so on ...
i then have to check what is the absolute maximum dimensions of driver i can use without creating beaming and other issues ...
so the drivers chosen are basically the smallest ones with the feature set i am looking for while also being the largest ones that will not cause acoustical issues in terms of dimensions being larger than a certain multiple of wavelength at the crossover frequency etc.
and of course i am trying to stick with reputable brands which generally means italian prosound OEMs but also JBL and Beyma ...
i use Radian and Aurum Cantus RELUCTANTLY because there are no equivalents from brands i respect ...
i then i have to go up 1 inch at a time and examine the options at every size level ...
generally the larger the driver the more advanced features you are able to have in it such as cast aluminum baskets, inside-outside voice coils, vented VC gap and so on ...
i then have to check what is the absolute maximum dimensions of driver i can use without creating beaming and other issues ...
so the drivers chosen are basically the smallest ones with the feature set i am looking for while also being the largest ones that will not cause acoustical issues in terms of dimensions being larger than a certain multiple of wavelength at the crossover frequency etc.
and of course i am trying to stick with reputable brands which generally means italian prosound OEMs but also JBL and Beyma ...
i use Radian and Aurum Cantus RELUCTANTLY because there are no equivalents from brands i respect ...
oh yeah the "wings" you see around the 8" Radian planar are a device to control diffraction.
without it a lot of output from the radian will go back and bounce off the main cabinet causing colorations.
but at the same time if i simply put the radian in a waveguide the output of 5" B&C mid behind it won't be able to get out ...
so what has to be done is a semi-permeable barrier made out of FOAM FINS ( made from Auralex foam ) that will act as an ACOUSTICAL LOW PASS FILTER and at the same time as RESONANCE CONTROL device ... it will allow low frequencies from 5" cone to get out while preventing the high frequencies from Radian from bouncing around too much in the space between it and the cabinet ...
i have used Auralex foam before it is amazing product. by controlling the size and shape of the fins and the gaps between them it should be possible to dial in the desired acoustical behavior of this semi-permeable barrier.
without it a lot of output from the radian will go back and bounce off the main cabinet causing colorations.
but at the same time if i simply put the radian in a waveguide the output of 5" B&C mid behind it won't be able to get out ...
so what has to be done is a semi-permeable barrier made out of FOAM FINS ( made from Auralex foam ) that will act as an ACOUSTICAL LOW PASS FILTER and at the same time as RESONANCE CONTROL device ... it will allow low frequencies from 5" cone to get out while preventing the high frequencies from Radian from bouncing around too much in the space between it and the cabinet ...
i have used Auralex foam before it is amazing product. by controlling the size and shape of the fins and the gaps between them it should be possible to dial in the desired acoustical behavior of this semi-permeable barrier.
furthermore the Aurum Cantus G3 ribbon would need to have its faceplate removed to reduce its width form 3" to 2" so that 4.5 khz from Radian can still go around it ...
G3 ribbon is 8.5 mm wide
by my estimates the aperture on 8" Radian is about 35 mm wide so about 4 X wider than G3 ribbon
this is why Radian begins to beam at around 7 khz but G3 has wide dispersion all the way up ...
some said you can just use Radian to 20 khz with no need for supertweeters ... and you certainly can ... and to be honest at my age i probably wouldn't even hear the difference ...
but let's face is Audiophile gear is all about impressing others. i can't just cut off the height of array based on my own height of 1 meter 80 cm and i can't just cut off HF response based on whatever frequency i can currently hear to ... no the speaker must be amazing to everybody who listens to it no matter their height or age ...
like why would i want to build a speaker that reminds everybody that i am not as tall or as young as perhaps i would like to be ? that would be dumb.
G3 ribbon is 8.5 mm wide
by my estimates the aperture on 8" Radian is about 35 mm wide so about 4 X wider than G3 ribbon
this is why Radian begins to beam at around 7 khz but G3 has wide dispersion all the way up ...
some said you can just use Radian to 20 khz with no need for supertweeters ... and you certainly can ... and to be honest at my age i probably wouldn't even hear the difference ...
but let's face is Audiophile gear is all about impressing others. i can't just cut off the height of array based on my own height of 1 meter 80 cm and i can't just cut off HF response based on whatever frequency i can currently hear to ... no the speaker must be amazing to everybody who listens to it no matter their height or age ...
like why would i want to build a speaker that reminds everybody that i am not as tall or as young as perhaps i would like to be ? that would be dumb.
slight reshuffling of components in the lower 3 frequency bands.
5" B&C 5NSM38 bucket is replaced with 18 Sound 6NMB420 midrange / midbass
https://www.eighteensound.it/en/products/lf-driver/6-5/8/6NMB420
i have developed a cooling solution that will be able to extract heat from the back chamber without needing a bucket midrange which can directly dissipate heat by having the chamber formed by the cast aluminum basket ... however based on hornresp "max spl" tool this heat exchanger i developed isn't really needed ... but i have that option as backup.
if anyone is interested my heat exchanger design is basically a way to sandwich 1/4" aluminum plate with 3/4" plywood in such a way that the resulting structure can form a non-resonant baffle that allows the air to flow between plywood and aluminum and expel heat through aluminum. JBL simply uses cast aluminum baffle on their VTX cabinets and my entire 5" JBL monitors are cast from aluminum, but for DIY purposes casting is not very practical. i then had to develop chamber geometry that allow the back wave to be damped with auralex foam without affecting the circulation of air that needs to pass over driver motor and aluminum plate. but now that i know it CAN be done ( though not necessary ) i can safely dispense with the 5NSM38 bucket mid and use real drivers like 6NMB420 instead.
Beyma 10NC700ND stays but i have also found a slightly lower cost ( $305 vs $365 ) alternative in B&C 10NDL88. Beyma has a pole vent and inside/outside VC while B&C has a larger VC that isn't any heavier because it's aluminum to Beyma's copper. They also pretty much match in things like FS, Xmax, MMS, BL, QTS etc. Beyma claims slightly higher AES power and slightly lower inductance and costs slightly more - that's about it. I would probably go with Beyma for the lower inductance ( even though it's only 20% lower ) and slightly more refined design but both would work and B&C does have cost advantage.
Beyond that after modeling the sub in HornResp and using Max SPL tool which models thermal and Xmax limitations and seeing that i am almost entirely displacement limited over anything else i decided to bump the 18" NTLW5000 to 21" NTLW5000
i find that Max SPL tool in HornResp is FAR more useful than simply modeling frequency response or even separately modeling frequency response and driver displacement. Because Max SPL tool instantly shows you whether you are more limited by Thermal or Mechanical at any given frequency - it's a brilliant tool.
so the current driver complement solution is:
4.5 + khz - Aurum Cantus G3
https://www.parts-express.com/Aurum...U9G9CQwlN8Zwa0M0iKVe1AxqbdajS2O9TM6JKvlL-WTcm
1.2 khz - 4.5 khz - Radian 8”
https://radianaudio.com/collections/ribbon/products/lm8k-wide-band-planar-ribbon-transducer
350 hz - 1.2 khz - 6NMB420 ( 3 per box )
https://www.eighteensound.it/en/products/lf-driver/6-5/8/6NMB420
80 hz - 350 hz - Beyma 10NC700ND or B&C 10NDL88 ( 2 per box )
https://www.beyma.com/speakers/Fich...rs-data-sheet-low-mid-frequency-10MC700Nd.pdf
https://www.bcspeakers.com/en/products/lf-driver/10/8/10NDL88
-80 hz -21” NTLW5000 ( 1 per box, asymmetrical vent )
https://www.eighteensound.it/en/products/lf-driver/21-0/4/21ntlw5000-4
5" B&C 5NSM38 bucket is replaced with 18 Sound 6NMB420 midrange / midbass
https://www.eighteensound.it/en/products/lf-driver/6-5/8/6NMB420
i have developed a cooling solution that will be able to extract heat from the back chamber without needing a bucket midrange which can directly dissipate heat by having the chamber formed by the cast aluminum basket ... however based on hornresp "max spl" tool this heat exchanger i developed isn't really needed ... but i have that option as backup.
if anyone is interested my heat exchanger design is basically a way to sandwich 1/4" aluminum plate with 3/4" plywood in such a way that the resulting structure can form a non-resonant baffle that allows the air to flow between plywood and aluminum and expel heat through aluminum. JBL simply uses cast aluminum baffle on their VTX cabinets and my entire 5" JBL monitors are cast from aluminum, but for DIY purposes casting is not very practical. i then had to develop chamber geometry that allow the back wave to be damped with auralex foam without affecting the circulation of air that needs to pass over driver motor and aluminum plate. but now that i know it CAN be done ( though not necessary ) i can safely dispense with the 5NSM38 bucket mid and use real drivers like 6NMB420 instead.
Beyma 10NC700ND stays but i have also found a slightly lower cost ( $305 vs $365 ) alternative in B&C 10NDL88. Beyma has a pole vent and inside/outside VC while B&C has a larger VC that isn't any heavier because it's aluminum to Beyma's copper. They also pretty much match in things like FS, Xmax, MMS, BL, QTS etc. Beyma claims slightly higher AES power and slightly lower inductance and costs slightly more - that's about it. I would probably go with Beyma for the lower inductance ( even though it's only 20% lower ) and slightly more refined design but both would work and B&C does have cost advantage.
Beyond that after modeling the sub in HornResp and using Max SPL tool which models thermal and Xmax limitations and seeing that i am almost entirely displacement limited over anything else i decided to bump the 18" NTLW5000 to 21" NTLW5000
i find that Max SPL tool in HornResp is FAR more useful than simply modeling frequency response or even separately modeling frequency response and driver displacement. Because Max SPL tool instantly shows you whether you are more limited by Thermal or Mechanical at any given frequency - it's a brilliant tool.
so the current driver complement solution is:
4.5 + khz - Aurum Cantus G3
https://www.parts-express.com/Aurum...U9G9CQwlN8Zwa0M0iKVe1AxqbdajS2O9TM6JKvlL-WTcm
1.2 khz - 4.5 khz - Radian 8”
https://radianaudio.com/collections/ribbon/products/lm8k-wide-band-planar-ribbon-transducer
350 hz - 1.2 khz - 6NMB420 ( 3 per box )
https://www.eighteensound.it/en/products/lf-driver/6-5/8/6NMB420
80 hz - 350 hz - Beyma 10NC700ND or B&C 10NDL88 ( 2 per box )
https://www.beyma.com/speakers/Fich...rs-data-sheet-low-mid-frequency-10MC700Nd.pdf
https://www.bcspeakers.com/en/products/lf-driver/10/8/10NDL88
-80 hz -21” NTLW5000 ( 1 per box, asymmetrical vent )
https://www.eighteensound.it/en/products/lf-driver/21-0/4/21ntlw5000-4
Last edited:
looks like i will be redesigning it again ...
started brushing up on line array theory as well as other things as well as modeling things in hornresp and the current design is not dialed in ...
i will need some time to figure a few things out before posting what will either be V3.5 or V4
basically i feel like the current design is bottlenecked in SPL by the Radian and the subs ... but i can't do anything about it ... the Radian is already the loudest driver of that type and so are the subs ...
what i really need to do is take down the output level between subs and the radian by a few decibels because in V3 it's getting wasted ...
the Ribbon is going to stay the same because as it turns out line arrays are actually most stressed in the upper register so i will need the loudest drivers there and Aurum Cantus is that.
i had previously considered Hi-Vi isodynamic ( planar ) supertweeter - that is now out of the question ! the output is not enough !
essentially line arrays naturally have a 3db / oct roll off across the entire operating range so the higher you go in frequency the louder drivers you need.
founteks are out as well, and so are the GRS. none of them are loud enough.
essentially the way line arrays work the supertweeter array will have to push the same SPL as the mid-tweeter array but do it at 1/4 the width of the ribbon. this will only be possible with pure aluminum true ribbon, so even fountek is out, let alone Hi-Vi.
GRS is out as well for a different reason - their ribbons are too short and flanges too wide. Aurum Cantus has longer ribbon with narrower flange, which is good for arraying.
so before doing any math my gut feeling here is i am about evenly matched on output between Aurum Cantus, Radian and NTLW subs but the 10" Beymas and 6" mids have too much output that the rest of the system won't keep up with. i will need to replace them with something better optimized for the application.
started brushing up on line array theory as well as other things as well as modeling things in hornresp and the current design is not dialed in ...
i will need some time to figure a few things out before posting what will either be V3.5 or V4
basically i feel like the current design is bottlenecked in SPL by the Radian and the subs ... but i can't do anything about it ... the Radian is already the loudest driver of that type and so are the subs ...
what i really need to do is take down the output level between subs and the radian by a few decibels because in V3 it's getting wasted ...
the Ribbon is going to stay the same because as it turns out line arrays are actually most stressed in the upper register so i will need the loudest drivers there and Aurum Cantus is that.
i had previously considered Hi-Vi isodynamic ( planar ) supertweeter - that is now out of the question ! the output is not enough !
essentially line arrays naturally have a 3db / oct roll off across the entire operating range so the higher you go in frequency the louder drivers you need.
founteks are out as well, and so are the GRS. none of them are loud enough.
essentially the way line arrays work the supertweeter array will have to push the same SPL as the mid-tweeter array but do it at 1/4 the width of the ribbon. this will only be possible with pure aluminum true ribbon, so even fountek is out, let alone Hi-Vi.
GRS is out as well for a different reason - their ribbons are too short and flanges too wide. Aurum Cantus has longer ribbon with narrower flange, which is good for arraying.
so before doing any math my gut feeling here is i am about evenly matched on output between Aurum Cantus, Radian and NTLW subs but the 10" Beymas and 6" mids have too much output that the rest of the system won't keep up with. i will need to replace them with something better optimized for the application.
My bet is you'll never finish this project, if you do, you won't keep it long.
I'd be delighted to be proven wrong.
I'd be delighted to be proven wrong.
I've lived with many big audio systems over the years, and as I grew older I came to realise a system that can do 120dB+ at the ears does little to improve the listening experience, both for music and movies. It also comes at a great cost, and I'm not just talking about financial cost.
The biggest advantage of those big systems of mine was the wow factor when I was selling products to customers. They were responsible for many large sales. I assume this is not beneficial to you?
The biggest advantage of those big systems of mine was the wow factor when I was selling products to customers. They were responsible for many large sales. I assume this is not beneficial to you?
yeah i'm not selling anything.
i dunno - i just can't design it any smaller. at least not the array.
i have the horn design that's slightly smaller ( with same SPL output ). i will post it later when i have the energy.
If i ever decide to spend the money to make a line array i would undoubtedly use compression drivers and line array waveguides. Its essentially the only way to eliminate comb filtering unless you use planar drivers.
ironically the version i'm currently mulling over is based on a line array of RCF NDT895 compression drivers to 9 khz then Ribbon above that
but the logic of using compression drivers over planars isn't to avoid comb filtering but to get more output
although planar waveguides like one you attached a picture of certainly vertically stack a lot closer than diaphragms of planars like Radian LM8K, but i was more worried about output capability of LM8K than closeness of stacking because the diaphragms on those are glued in place and can't be replaced ...
although a diaphragm for the RCF is almost as much as the whole LM8K driver the RCFs are proven workhorses and i don't expect them to ALL die for no good reason - if one or two has a diaphragm failure those can be replaced