Thanks to all the amazing help I got in my first thread musing about flexible PCBs (and why not to do it) I decided to scrap my requirements and start again.
My new requirements are:
Basically I want the perfect purpose build planar driver for my new dipole CBT which will look very similar to the old one
but with a planar as the main driver instead of full range 2.5"cone drivers. I'll still use cone drivers for 80-400 hz but I'm planning to hide them behind the planar in the spine of the speaker.
I will at least initially use 12 micron mylar & 30 micron aluminum foil. I will cut the foil on a silhouette cameo 4. I think I will need ~ 1 mm clearance between traces of the coil but I'll have to test that when I get parts & the machine. But since others here on the forum has done this already I'm not that worried and it will probably work just fine.
I also have plans to Corrugate the membrane lengtwise like this to stiffen side-to-side resonances. And then mount it top and bottom while floating (with foam) at the sides to reduce resonance frequency.
There are lots of small desicions I need to but for now the next open question is:
The 3 lengthwise setups are pretty normal. Just a question of how big the gap between the magnets can be without it impacting performance. My gut says that of course the 7 magnet setup will measure better in the treble but how much better and would it be significant? Having only 3 rows of larger magnets would probably be much easier to build. Since I wont have a metal front plate I'd have to get creative making a strong enough frame that I can print. But should be solvable although I might need to add more frame around the magnets to help support them.
The magnet-sideways mounted setup is kinda weird but would have some distinct advantages: It would be easier for me to manifacture on my 3d printer while at the same time having less support since I can rely on the magnets themselves for mechanical support. It would also probably measure extremely smooth in the vertical axis since it is extremely uniform thus should have an effective minimal CTC.
It does have some glaring disadvantages though: I'd have to come with some way to ensure that all the magnets and the coils are in alignment which isn't a trivial problem to fix. I'd also have to get creative when corrugating the membrane since all the corrugations would need to be uniform along the whole length. Also, it would be slightly less efficient due to the coil twisting more. It would, however, be fun to try to overcome the engineering challenges and build it anyway
First is waiting for materials but then plan to build and test prototypes. My plan is to build 20 cm tall prototypes since the main thing I want to test is how the different setups perform in the horizontal axis + how easy they are to manifacture and how mechanically stable they are. Or basically how much extra support do I have to add to make it not fall apart, less is better.
The design which measures good enough while also being easy enough to manifacture will be the winner
My new requirements are:
- Single planar driver, 160 cm tall.
- Good frequency response, no ugly dipole peak allowed.
- Physically curved with a 75 cm radius.
- Shadeable, ideally as an internal part of the coils but splitting the coils into multiple separate parts that can be driven separately would be OK.
- Can cover from 300-400 hz (24 dB / oct crossover) up to 20 khz.
- Can output 100 dB @ 1m across the operating frequency range without too much distortion.
- Comparable horizontal dispersion to a Neo3W, limits horizontal radiating area to ~ 35-40 mm wide in the treble.
- Should behave as a single driver with vertical CTC of at most 2 cm. I want to avoid vertical lobing.
- Manifacturable with a 3d printer + using laser cut plywood.
Basically I want the perfect purpose build planar driver for my new dipole CBT which will look very similar to the old one
but with a planar as the main driver instead of full range 2.5"cone drivers. I'll still use cone drivers for 80-400 hz but I'm planning to hide them behind the planar in the spine of the speaker.I will at least initially use 12 micron mylar & 30 micron aluminum foil. I will cut the foil on a silhouette cameo 4. I think I will need ~ 1 mm clearance between traces of the coil but I'll have to test that when I get parts & the machine. But since others here on the forum has done this already I'm not that worried and it will probably work just fine.
I also have plans to Corrugate the membrane lengtwise like this to stiffen side-to-side resonances. And then mount it top and bottom while floating (with foam) at the sides to reduce resonance frequency.
There are lots of small desicions I need to but for now the next open question is:
How should I setup my magnets?
I'm tossing around ideas for 4 configurations. Since I like tinkering I might just build all 4 and then measure them and see which one I like the best.The 3 lengthwise setups are pretty normal. Just a question of how big the gap between the magnets can be without it impacting performance. My gut says that of course the 7 magnet setup will measure better in the treble but how much better and would it be significant? Having only 3 rows of larger magnets would probably be much easier to build. Since I wont have a metal front plate I'd have to get creative making a strong enough frame that I can print. But should be solvable although I might need to add more frame around the magnets to help support them.
The magnet-sideways mounted setup is kinda weird but would have some distinct advantages: It would be easier for me to manifacture on my 3d printer while at the same time having less support since I can rely on the magnets themselves for mechanical support. It would also probably measure extremely smooth in the vertical axis since it is extremely uniform thus should have an effective minimal CTC.
It does have some glaring disadvantages though: I'd have to come with some way to ensure that all the magnets and the coils are in alignment which isn't a trivial problem to fix. I'd also have to get creative when corrugating the membrane since all the corrugations would need to be uniform along the whole length. Also, it would be slightly less efficient due to the coil twisting more. It would, however, be fun to try to overcome the engineering challenges and build it anyway
Next step
First is waiting for materials but then plan to build and test prototypes. My plan is to build 20 cm tall prototypes since the main thing I want to test is how the different setups perform in the horizontal axis + how easy they are to manifacture and how mechanically stable they are. Or basically how much extra support do I have to add to make it not fall apart, less is better.
The design which measures good enough while also being easy enough to manifacture will be the winner
Last edited:
The point is that this type of arrays are used in professional sound engineering, where the task is to bring sound to the front and back rows of listeners in large halls.
But in domestic use the sound from the upper part of the curved driver will go past the listener to the ceiling, this in turn will cause cutoff and interference mush, that is, will be brightly colored.
But in domestic use the sound from the upper part of the curved driver will go past the listener to the ceiling, this in turn will cause cutoff and interference mush, that is, will be brightly colored.
Attachments
The top of the array is shaded. The end result is that the vertical dispersion is reduced and controlled.
In practice it measures like a floor to ceiling array regardless of ceiling height. I also think the physically curved design looks really nice and not as bulky as if it was straight and taller.
Both, kinda.
I want to build a single continously curved driver, with a single membrane but to make manifacturing easier I plan to build the frames into 8 segments @ 20 cm each. I think that will work since the membrane won't be tensioned like the usual planars. I will have a more robust plywood frame which will hold all the segments and keep them in place.
Not planning on doing any magnet analysis no. I planned on just copying a known design rather than trying to get creative with the magnet structure. I plan to have +- 1 mm of mechanical xmax + the peak to peak height of the corrugations + some appropriate margin so that I don't hit the magnets if everything isn't perfect everywhere. And then just fit as many magnets as I can with appropriate size without penalizing the frequency response too much. The current plan is to use tiny 3x3 mm magnets to ensure that the frequency response is smooth.
Ok, that's smart. So you could in theory make an ordinary line array if you are not satisfied with the CBT.
I mean, the shading will be all about the foil topology.
Will it be a single ended or a push-pull magnet structure?
Looks like that you will need a somewhat even magnetic field density of almost +/- 4 mm in total.
Are the magnets 3x3x3 mm?
I mean, the shading will be all about the foil topology.
Will it be a single ended or a push-pull magnet structure?
Looks like that you will need a somewhat even magnetic field density of almost +/- 4 mm in total.
Are the magnets 3x3x3 mm?
The magnets I've bought are 40x5x5, 50x3x3 N35 (and some N52 although I'm expecting it to be "N52"), I don't know what dimension I'll use in the end although I'm leaning towards the 3x3 since they would obstruct less. But this is just what I bought just now to play around with, no huge amounts so if I find other dimensions work better then I'll go with that instead.
The current plan is to do a push pull since I want to cross it low, hence why I'm extra interested in not having the magnets be too big. Also same reason I'm still thinking about at least testing the sideways mounted magnet setup because then I don't need a pole piece which in the end means less obstructions for the sound and in theory more smooth response.
And as long as I have in +- 0.1 mm xmax with minimal distortion so I can cross 400 hz @ 24 dB then the base goal of the driver is met. If I can push it to up to say +- 0.25-0.5 mm it would probably be even better.
The current plan is to do a push pull since I want to cross it low, hence why I'm extra interested in not having the magnets be too big. Also same reason I'm still thinking about at least testing the sideways mounted magnet setup because then I don't need a pole piece which in the end means less obstructions for the sound and in theory more smooth response.
And as long as I have in +- 0.1 mm xmax with minimal distortion so I can cross 400 hz @ 24 dB then the base goal of the driver is met. If I can push it to up to say +- 0.25-0.5 mm it would probably be even better.
Yes, I think 3x3 mm is the better choice.
If you're doing push pull, you'll have to consider the Helmholtz cavity resonance frequency and not so much the obstruction of the reproduced sound wave.
Mounting the magnets horizontally and thus not having any pole pieces will for sure help, but the magnetic field density will be more un-even and the unsupported magnets can be a problem being fragile.
(Note that I didn't mentioned FEMM analysis. Darn, now I did it!)
Sorry, then I missunderstod your post above.
If you're doing push pull, you'll have to consider the Helmholtz cavity resonance frequency and not so much the obstruction of the reproduced sound wave.
Mounting the magnets horizontally and thus not having any pole pieces will for sure help, but the magnetic field density will be more un-even and the unsupported magnets can be a problem being fragile.
(Note that I didn't mentioned FEMM analysis. Darn, now I did it!)
Sorry, then I missunderstod your post above.
With +/- 0.1 mm and even +/- 0.5 mm you'll do fine.
This is the latest idea if I go with sideways mounted magnets:
The wavyness in the lines would ensure that sideways line of the membrane is unsupported without any aluminium traces on it. And in theory, if I can focus the corrugations in the gap as I mused earlier then this design should get away with the magnets being closer than a normal lengthwise magnet setup. So we gain some efficiency but we loose 25% to all the extra loops so in the end the straight will probably be more efficient.
If I go with straight magnets then I can always use perforated steel plates but they obstruct too much for my liking and would be annoying to work with. So I threw together a rough sketch 200x70x1.2mm part which assumes 5 rows of 50x3x3 magnets + some screw holes.
Then sent it to some online laser cut services to see if it would be cheap enough to use. And it would, if I cut 50 of those parts which would be enough for both speakers it would cost €220 including shipping from https://fractory.com/sv/laserskarning/ The previous manifacturer I checked at (https://cotter.se/) is extremely expensive for steel, a single (1!) part would with shipping would cost more than that...
Although that is with 50x of a single part, I'd need 20x the outer part and 20x the inner part since the outer part will be just slightly longer since the radius will be bigger.
But with laser cut parts then they will be consistent enough that it should be simple to print extra helper parts like press moulds to ensure they all have the appropriate curve and jigs to ensure all the magnets are glued in the correct place for all the segments. Also, if I go with normally aligned magnets like in the metal cut part then I'll probably see if I can buy shorter magnets, say 30x3x3 mm magnets make them fit in the curve with less height error in the middle.
The wavyness in the lines would ensure that sideways line of the membrane is unsupported without any aluminium traces on it. And in theory, if I can focus the corrugations in the gap as I mused earlier then this design should get away with the magnets being closer than a normal lengthwise magnet setup. So we gain some efficiency but we loose 25% to all the extra loops so in the end the straight will probably be more efficient.
If I go with straight magnets then I can always use perforated steel plates but they obstruct too much for my liking and would be annoying to work with. So I threw together a rough sketch 200x70x1.2mm part which assumes 5 rows of 50x3x3 magnets + some screw holes.
Then sent it to some online laser cut services to see if it would be cheap enough to use. And it would, if I cut 50 of those parts which would be enough for both speakers it would cost €220 including shipping from https://fractory.com/sv/laserskarning/ The previous manifacturer I checked at (https://cotter.se/) is extremely expensive for steel, a single (1!) part would with shipping would cost more than that...
Although that is with 50x of a single part, I'd need 20x the outer part and 20x the inner part since the outer part will be just slightly longer since the radius will be bigger.
But with laser cut parts then they will be consistent enough that it should be simple to print extra helper parts like press moulds to ensure they all have the appropriate curve and jigs to ensure all the magnets are glued in the correct place for all the segments. Also, if I go with normally aligned magnets like in the metal cut part then I'll probably see if I can buy shorter magnets, say 30x3x3 mm magnets make them fit in the curve with less height error in the middle.
Last edited:
In that case, corrugated along the sideways mounted magnets. And yes you are correct.
Maybe you could order 50 and sand/grind off the mm that need to go off on 20 of them?
Please do a complete drawing - I'm curious of how it is going to be done and it is always a little hard to imagine these by just words...
Sanding would be way too much work, and the main appeal of laser cutting is that every dimension would be exactly right. Far beyond what I can produce if I adjust too much manually. I threw up another instant quote in https://fractory.com but with slightly different parts with 20 parts each. The price is bumped by 10 € which is nice and cheap.
And the plan for the weekend is to make a CAD model of the straight magnet variant
.I've been thinking more about what you wrote @solhaga about magnets & push pull vs single ended.
My main goal is that I want great and smooth frequency response on the front and SE would be better at that, it would solve the cavity resonance problem too. And the rear should still be good enough such that it measures well enough as a dipole.
By going SE instead of PP the main result should be less even harmonics cancellation right? I'm OK with more 2nd, if anything I've found I prefer it to 3rd when toying with SE class A flea amps like the Aleph J. I could add more / bigger magnets on the rear to compensate for the reduced field strength and it would be slightly easier to manifacture. I would only need 1 magneted side and wouldn't need to force two magneted sides together risking ripping the membrane
My main goal is that I want great and smooth frequency response on the front and SE would be better at that, it would solve the cavity resonance problem too. And the rear should still be good enough such that it measures well enough as a dipole.
By going SE instead of PP the main result should be less even harmonics cancellation right? I'm OK with more 2nd, if anything I've found I prefer it to 3rd when toying with SE class A flea amps like the Aleph J. I could add more / bigger magnets on the rear to compensate for the reduced field strength and it would be slightly easier to manifacture. I would only need 1 magneted side and wouldn't need to force two magneted sides together risking ripping the membrane
Drawings!
Currently assuming I use 12x5x3 mm magnets, 17 lengthwise. Driver is 7 cm wide and has a 75 cm radius. The yellow parts will be printed in PETG (or PETG-CF) on my printer, the gaskets will be printed in porous Varioshore TPU in 65A shore firmness. The top and bottom plates will be 1.5 mm steel laser cut. Not yet modeled is that I plan to glue a thin non woven fleece fabric on top of the magnets to reduse resonances. The round thingys on the 3d printed parts and the gaskets are just to keep them in alignment without glue. The only glue I'm planning for is for the magnets on the steel plates.
Note that the driver is open on the top and bottom. The idea is to connect 8x of these on top of each other with a common membrane. I'll probably print small specialized bottom and top parts where I attach the membrane.
Open questions. If I can't reason myself to the answers I think I'll have to try out the different options.
Currently assuming I use 12x5x3 mm magnets, 17 lengthwise. Driver is 7 cm wide and has a 75 cm radius. The yellow parts will be printed in PETG (or PETG-CF) on my printer, the gaskets will be printed in porous Varioshore TPU in 65A shore firmness. The top and bottom plates will be 1.5 mm steel laser cut. Not yet modeled is that I plan to glue a thin non woven fleece fabric on top of the magnets to reduse resonances. The round thingys on the 3d printed parts and the gaskets are just to keep them in alignment without glue. The only glue I'm planning for is for the magnets on the steel plates.
Note that the driver is open on the top and bottom. The idea is to connect 8x of these on top of each other with a common membrane. I'll probably print small specialized bottom and top parts where I attach the membrane.
Open questions. If I can't reason myself to the answers I think I'll have to try out the different options.
- Should the corrugations really be that deep? Also, I think I'll need to make my corrugator myself since I need to have the corrugations match the gaskets.
- How much distance should I have from the top of the membrane to the magnets?
- How tight should the gaskets be against the membrane? Maybe they should be pretty loose and just roughly hold it in alignment?
- Should I print the gaskets solid? The material should be pretty porous but I could print with infill to make even softer or skeletonize it to let the membrane move easier.
- Is the small area where the membrane connects to the gasket enough to support it or should it be longer?
- Is 1.5 mm steel appropriate for the top and bottom, should I go with thicker like 2 mm?
- Are 5x3 mm magnets appropriate? I have 4x3 mm magnets ordered too which I can also try. I guess I also have 3x3 but those are longer so will not keep the curve as well as the smaller so I might try them in a flat setup.
- Should I have magnets on both sides? This should be easy to test since I can just measure the driver again without mounting the front plate & front magnets.
Attachments
Last edited:
Here's some thoughts.
1. I'm thinking +/- 0.5 mm is enough, that is 1 mm in total. But you can corrugate it "deeper" and then stretch the membran to the right depth.
And why don't you design the gasket's corrugation after the corrugation of the membrane has been decided?
2. I think you need to do some FEMM analysis of that.
3. How long will the gasket/dampening zone be? Perhaps the membrane can be a fairly tight at the end and then there's a section of looser and looser dampening (foam).
4. If you'll print it with the same filament, you can have it firmer at the end and then looser and looser towards the active part of the membrane.
5. I don't think so. But you might want a longer dampening zone.
6. Will the steel plate come curved from the manufacturer? How does the magnets attach to the curved surface? I don't know if it should be 1.5 mm or 2.0 mm, you'll have to test how much the bulge is in the middle.
7. I think you need to do some FEMM analysis of that.
8. Upside of a push/pull configuration is that you'll get a more even magnetic field density over the excursion of the membrane. Downside is a more complicated motor to assemble and that you have to consider the Helmholtz' resonators. Here's a calculator.
1. I'm thinking +/- 0.5 mm is enough, that is 1 mm in total. But you can corrugate it "deeper" and then stretch the membran to the right depth.
And why don't you design the gasket's corrugation after the corrugation of the membrane has been decided?
2. I think you need to do some FEMM analysis of that.
3. How long will the gasket/dampening zone be? Perhaps the membrane can be a fairly tight at the end and then there's a section of looser and looser dampening (foam).
4. If you'll print it with the same filament, you can have it firmer at the end and then looser and looser towards the active part of the membrane.
5. I don't think so. But you might want a longer dampening zone.
6. Will the steel plate come curved from the manufacturer? How does the magnets attach to the curved surface? I don't know if it should be 1.5 mm or 2.0 mm, you'll have to test how much the bulge is in the middle.
7. I think you need to do some FEMM analysis of that.
8. Upside of a push/pull configuration is that you'll get a more even magnetic field density over the excursion of the membrane. Downside is a more complicated motor to assemble and that you have to consider the Helmholtz' resonators. Here's a calculator.
1. I will. The whole drawing is parametrized so my next step is to start actually corrugating some example membranes and see what shape it actually gets and then correct the drawing.
2. Sounds reasonable. I tried to plot in https://www.femm.info/wiki/HomePage but the UI wasn't the most intuitive so for that time I gave up after a few tries. I might have to try again or to just print parts for multiple configurations and see in practice. If I want to test magnet distance I might tweak the curve such that the driver is flat because then I can adjust the distance without having to redo the sheet metal parts with the magnets.
3. I wonder that too . When I have a prototype I might as well print all 3 combinations and then measure them all. Might also be worth trying to print the gasket in one piece on each side without cutting it in the middle. Might also be worth narrowing down the gasket under the magnets to let it move more.
4. True, but would make it more complicated to print. Probably easier to play around with more or less infill and more or less supported to achieve that.
5. With dampening zone you mean the overlap of the gasket and the membrane?
6. I hope that I can order them curved from the manifacturer yes but I haven't found yet how I can supply the bend metadata such that they understand it to be a 75 cm radi bend. If I don't find out on my own I'll probably contact them and ask them, they should have the tools for it. Otherwise I'll have to bend them myself but I'd rather avoid that. The Neo3 seems to use 1.5 mm steel but that is pressed on the sides and top which stiffens it a lot. This one will be curved though which will make it more stiff.
On the topic of the magnets: The plan is to glue them to the steel plate with 24 hour epoxy glue. And to make sure they are all glued perfectly I plan to print an inside of the speaker filled up except for voids where the magnets are. So I can add glue to all the magnets and roughly place them, then screw the plate into the magnet holding frame which forces all magnets to go exactly where they should and then leave it to dry. I might need to coat the holding frame to ensure it doesn't stick to the glue but apart from that I believe it will let me glue all the magnets with great accuracy and consistency.
8. If I plug the current cavity size into the calculator I get 9.5 khz resonance which is close enough to 10k. And it all depends on how sharp the resonance peak is, if it is gentle I can just EQ it away.
2. Sounds reasonable. I tried to plot in https://www.femm.info/wiki/HomePage but the UI wasn't the most intuitive so for that time I gave up after a few tries. I might have to try again or to just print parts for multiple configurations and see in practice. If I want to test magnet distance I might tweak the curve such that the driver is flat because then I can adjust the distance without having to redo the sheet metal parts with the magnets.
3. I wonder that too
. When I have a prototype I might as well print all 3 combinations and then measure them all. Might also be worth trying to print the gasket in one piece on each side without cutting it in the middle. Might also be worth narrowing down the gasket under the magnets to let it move more.4. True, but would make it more complicated to print. Probably easier to play around with more or less infill and more or less supported to achieve that.
5. With dampening zone you mean the overlap of the gasket and the membrane?
6. I hope that I can order them curved from the manifacturer yes but I haven't found yet how I can supply the bend metadata such that they understand it to be a 75 cm radi bend. If I don't find out on my own I'll probably contact them and ask them, they should have the tools for it. Otherwise I'll have to bend them myself but I'd rather avoid that. The Neo3 seems to use 1.5 mm steel but that is pressed on the sides and top which stiffens it a lot. This one will be curved though which will make it more stiff.
On the topic of the magnets: The plan is to glue them to the steel plate with 24 hour epoxy glue. And to make sure they are all glued perfectly I plan to print an inside of the speaker filled up except for voids where the magnets are. So I can add glue to all the magnets and roughly place them, then screw the plate into the magnet holding frame which forces all magnets to go exactly where they should and then leave it to dry. I might need to coat the holding frame to ensure it doesn't stick to the glue but apart from that I believe it will let me glue all the magnets with great accuracy and consistency.
8. If I plug the current cavity size into the calculator I get 9.5 khz resonance which is close enough to 10k. And it all depends on how sharp the resonance peak is, if it is gentle I can just EQ it away.
Attachments
If the gray part is the gasket I guess that you don't have any dampening zone, just a little less gasket towards the active parts of the membrane.
How do you get the membrane into the gasket?
So the magnets's surfaces will barely touch the bent metall?
10 kHz, that's about what I got as well for my contraption.
How do you get the membrane into the gasket?
So the magnets's surfaces will barely touch the bent metall?
10 kHz, that's about what I got as well for my contraption.
the dark gray part is the gasket, which is light green in this image.
And how I insert the membrane in the gasket: If I do a split gasket then I guess half is placed on the bottom of the driver, then the membrane and then the top and sandwich it together with the screws on the side. It I print them in one piece then I guess I'll slide the membrane into the gaskets first and then place the gasket + membrane combo into the rest of the driver and then sandwich it together.
In the current plans then no, they will not. Is that a problem?
Attachments
Interesting! Does that mean that the glue is usually just on the sides of the magnets then for maximum performance?
I guess I could do a stepped curve in a single sheet of steel but that sounds really complicated. The manifacturer might be able to do it but the tolerances would have to be amazing so probably expensive
. The simplest way to solve that issue is probably to remove the curve on the steel plates and make them flat. But also split it into many parts where each part might be say 12-50 mm long. Would it be a problem if the small steel segments are not directly connected to each other?Then get the curve through the middle parts which is 3d printed. That might be a big weak though so it is probably a good idea if I can cram some steel reinforcements in the yellow 3d printed parts that are curved but cover the whole arc of each ~ 20 cm long driver segment. Or since it is already split I might as well have these steel reinforcements cover the whole 160 cm tall driver.
Attachments
Last edited: