Howdy folks. I'm interested in experimenting with 3d printing technology to create fancy horns.
I currently have a Reprap Prusa 3D Printer, with a build area of about 190x190x100mm
The nice thing about the 3d printer is it's ability to handle complex shapes just as easily as very simple shapes. It just handles any 3d model 1 layer at a time, so you can make some pretty wild or intricate designs.
I was thinking a fun thing to try at first, and given my relatively small build area, would be some sort of phone dock with integrated horn/waveguide for amplifying a cell phone speaker. I realize this won't be winning any kind of awards for sound quality, but I'm interested in understanding the concepts, and seeing how much of an effect it can have. And maybe in the future using this knowledge I could scale up the idea for a full range enclosure with a serious driver in it.
So I'm wondering if anyone can provide some tips for how to design a horn profile for maximum efficiency and sound quality. Would I have to know very specific specifications(thiele&small?) of the actual driver used in the phone? I have an HTC EVO which I would be using to test this with. Is there some way I could measure or derive whichever significant parameters I would need for this?
One thing I'm curious about is if there is an optimal total length of the horn, or if you can just keep going larger and larger for more and more effect.
Here is an example of a commercial product produced for the iPhone which is the sort of thing I would be interested in doing:
AirCurve - fits iPhone 3G and original iPhone - Griffin Technology
I currently have a Reprap Prusa 3D Printer, with a build area of about 190x190x100mm
The nice thing about the 3d printer is it's ability to handle complex shapes just as easily as very simple shapes. It just handles any 3d model 1 layer at a time, so you can make some pretty wild or intricate designs.
I was thinking a fun thing to try at first, and given my relatively small build area, would be some sort of phone dock with integrated horn/waveguide for amplifying a cell phone speaker. I realize this won't be winning any kind of awards for sound quality, but I'm interested in understanding the concepts, and seeing how much of an effect it can have. And maybe in the future using this knowledge I could scale up the idea for a full range enclosure with a serious driver in it.
So I'm wondering if anyone can provide some tips for how to design a horn profile for maximum efficiency and sound quality. Would I have to know very specific specifications(thiele&small?) of the actual driver used in the phone? I have an HTC EVO which I would be using to test this with. Is there some way I could measure or derive whichever significant parameters I would need for this?
One thing I'm curious about is if there is an optimal total length of the horn, or if you can just keep going larger and larger for more and more effect.
Here is an example of a commercial product produced for the iPhone which is the sort of thing I would be interested in doing:
AirCurve - fits iPhone 3G and original iPhone - Griffin Technology
Hmmm, indeed... I would venture a guess that many if not most folks here aspire to projects on a slightly larger scale. Not to dissuade you - I myself have some tiny (~1") speakers salvaged from a laptop PC just waiting for horns something along the lines of what you're suggesting. I've seen a variety of creative efforts to accomplish this sort of thing. Some of them are relatively crude, cardboard and masking tape affairs, yet they manage some impressive results. Maybe not hi-fi, but still a useful improvement. I've even come across plans for a matchbox sized, kardboard Karlson enklosure!
Design tips... lets see:
First off, don't be intimidated by the "mathematically engineered" design in your linked example. I believe that engineering of interesting speakers - including folded/convoluted horns - can't be distilled down to just mathematics. As for measuring parameters of the phone's speaker driver, that may be of use eventually in refining your design, but why not start with some experimental builds based on the dimensional constraints at hand? Perhaps you could construct some straight (non-convoluted) cardboard/tape mockups to evaluate different horn expansion profiles and mouth areas before consuming a lot of - I imagine - pricey printer ammo. Listening tests at this stage would probably be the most useful indicator of whether the concept has the potential to perform to your expectations.
I'd suggest a throat section that tightly couples to the speaker opening on your phone - try to minimize discontinuities or cavities that would surely lead to troublesome resonances. The (eventually) convoluted section should be of narrow, minimally expanding cross-section with curve radii as large as possible relative to cross-section diameter. Once you get it to the desired direction, then you can do the final expansion to the largest practical mouth opening. Expansion profile seems to elicit a lot of emotional discussion - I guess it can affect things like directivity. Once again, some quick mock-ups should give you some idea of how critical that decision will be, and how to direct your further efforts. A conical expansion is pretty hard to beat for simplicity, and personally, I'd start with that.
Once you've made a start and have some initial impressions (and pictures!) to share, I'm sure more pointed suggestions will follow.
By the way, I'm curious about your CAD software of choice for your printer modeling - care to share?
Wilf
Last edited:
re: Karlson-coupler - here's approximately the size for you already done 50 years ago 
http://vintageradio.me.uk/radconnav/comp&data/speakercabs/page1.jpg
http://vintageradio.me.uk/radconnav/comp&data/speakercabs/page2.jpg
http://vintageradio.me.uk/radconnav/comp&data/speakercabs/page1.jpg
http://vintageradio.me.uk/radconnav/comp&data/speakercabs/page2.jpg
You called?
http://www.diyaudio.com/forums/full...ng-little-different-quick-easy-ipod-horn.html
Chris
PS - I recently got a new phone, one that actually plays MP3s (had an old Nokia brick before), so stay tuned for another cardboard horn.
Last edited:
Thanks for the links and great ideas everyone.
Another option which is popular with other 3d printing people is OpenSCAD. It has it's own scripting language, which is fairly simple and easy to learn, but not nearly as powerful as Python.
I have been using mostly FreeCAD lately. I am a software developer by trade, and I prefer to create models through scripted CAD. This allows for parametric designs, where you can change a variable and see the output without redesigning an entire part. FreeCAD uses Python for it's scripting, which is one of my more preferred languages.
Another option which is popular with other 3d printing people is OpenSCAD. It has it's own scripting language, which is fairly simple and easy to learn, but not nearly as powerful as Python.
Well, I know it was suggested that I make some stuff with cardboard first, but generating 3D models is just too much fun.
I've put together a script for generating a wide variety of horn shapes. It is written in Python, for FreeCAD. It mostly generates the Mesh vertexes independently of the FreeCAD functions, so it could possibly be split out to a standalone script. I'm mainly using FreeCAD for visualization and ability to export to common 3d formats in this case.
The idea is that you can provide mathematical functions for each of:
Some examples just for fun:
Radius functoin: Oblate Spheroid
Path Function: Linear
Section: circular
A bunch of exponential radius horns, linear path, circular cross section
exponentinal "radius", circular path along 3/4 turns, Rectangular cross section
I'm still unsure about the way to define certain horn radius equations though. For example exponential horns. I know the simplest form of exponential equation is y = e^x. But this doesn't give any parameters for starting radius, or flare rate(is that the correct term?). Is there a common form of this function used for making horn profiles? I've implemented a sort of exponential function already, just not sure if I did it "properly".
I've also read about tractrix as another common profile, but I'm having trouble finding any euqation to model the radius for that as well.
Plus any other common shapes that I might be missing. Is there any cheatsheet of common horn equations on these boards or elsewhere online?
The script is still a work in progress, but I will be releasing it as open source in the near future when I feel it is more polished.
I've put together a script for generating a wide variety of horn shapes. It is written in Python, for FreeCAD. It mostly generates the Mesh vertexes independently of the FreeCAD functions, so it could possibly be split out to a standalone script. I'm mainly using FreeCAD for visualization and ability to export to common 3d formats in this case.
The idea is that you can provide mathematical functions for each of:
- The horn path, which is the line that the center of the horn follows in 3d space, so you can have straight horn or spiral shapes, etc.
- The radius as a function of distance along the horn path
- The horn cross section(circular or rectangular for example), as a function of the radius
Some examples just for fun:
Radius functoin: Oblate Spheroid
Path Function: Linear
Section: circular
A bunch of exponential radius horns, linear path, circular cross section
exponentinal "radius", circular path along 3/4 turns, Rectangular cross section
I'm still unsure about the way to define certain horn radius equations though. For example exponential horns. I know the simplest form of exponential equation is y = e^x. But this doesn't give any parameters for starting radius, or flare rate(is that the correct term?). Is there a common form of this function used for making horn profiles? I've implemented a sort of exponential function already, just not sure if I did it "properly".
I've also read about tractrix as another common profile, but I'm having trouble finding any euqation to model the radius for that as well.
Plus any other common shapes that I might be missing. Is there any cheatsheet of common horn equations on these boards or elsewhere online?
The script is still a work in progress, but I will be releasing it as open source in the near future when I feel it is more polished.
OK, I found this paper about exponential horn which explains the equation in the first page and a half. http://www.google.com/url?sa=t&rct=...DvSgUEyK5CJlgX-Iw&sig2=2yq6jXjoXfpx404rKHCuGQ
Cool, so now I can just specify a start radius, end radius, and length, and it computes the rest.
Still trying to figure out the tractrix, and maybe hyerbolic, or others.
Cool, so now I can just specify a start radius, end radius, and length, and it computes the rest.
Still trying to figure out the tractrix, and maybe hyerbolic, or others.
Well, I printed out a test horn. It is exponential with a throat diameter of 10mm and mouth diameter of 100mm. The length is 100mm.
It definitely amplifies the sound, but mostly the higher frequencies. I am now planning to make some sort of folded design so that I can greatly increase the length of the wavepath, since 100mm is about as tall as I can print.
I read from Martin King's paper here about an equation for lower cutoff frequency based on the flare constant m. http://www.quarter-wave.com/Horns/Horn_Physics.pdf
He gives these equations
m = ln(Sl / S0) / L
fc = (m * c) / (4pi)
And using the parameters for my printed horn, I calculate a lower cutoff frequency of 1247Hz.
But then further down in the same paper, there is a Table 5.2 where the frequencies are calculated completely differently. It says "In Table 5.2, the lower cut-off frequency
fc is calculated based on the open end (the horn’s mouth) cross-sectional area." But I don't know what equation is used for that, and it seems to have opposite predictions from the first cutoff frequency equations that he gives for exponential horn. i.e. if you have a constant throat area and length, and you increase the mouth area, the first equations seem to indicate that your cutoff frequency would rise since the constant m would be higher. However, this table 5.2 shows the cutoff frequency dropping as the mouth is widened.
Can anyone explain that?
Edit: I uploaded a video of it in action http://youtu.be/85x4PlhWHkc
It definitely amplifies the sound, but mostly the higher frequencies. I am now planning to make some sort of folded design so that I can greatly increase the length of the wavepath, since 100mm is about as tall as I can print.
I read from Martin King's paper here about an equation for lower cutoff frequency based on the flare constant m. http://www.quarter-wave.com/Horns/Horn_Physics.pdf
He gives these equations
m = ln(Sl / S0) / L
fc = (m * c) / (4pi)
And using the parameters for my printed horn, I calculate a lower cutoff frequency of 1247Hz.
But then further down in the same paper, there is a Table 5.2 where the frequencies are calculated completely differently. It says "In Table 5.2, the lower cut-off frequency
fc is calculated based on the open end (the horn’s mouth) cross-sectional area." But I don't know what equation is used for that, and it seems to have opposite predictions from the first cutoff frequency equations that he gives for exponential horn. i.e. if you have a constant throat area and length, and you increase the mouth area, the first equations seem to indicate that your cutoff frequency would rise since the constant m would be higher. However, this table 5.2 shows the cutoff frequency dropping as the mouth is widened.
Can anyone explain that?
Edit: I uploaded a video of it in action http://youtu.be/85x4PlhWHkc
Last edited:
Hi:
Very nice effort - thanks for the picture and video.
May I suggest including a flange at the base - it would make it easier to balance it on your phone, plus it would assist sealing the joint with blu-tac or such, as mentioned. Sounds like you're on to bigger models - again, flanges at the mating joints should be helpful with alignment and connection of segments (whether the axis is straight or curved).
Sorry, I can't help you with the finer points of horn theory w.r.t. various profiles, cutoff frequency, etc. As I said though, the differences may not be hugely significant in your application - I'm sure that little speaker has a very limited low-frequency output, so I don't think any horn size or shape will drastically change that. But hey, have fun experimenting! Is the small end opening exactly matched to the speaker opening? Is the speaker sound emanating from a single opening, or is it an array of tiny holes? As I said, creating a step-like change in cross-section at that juncture would likely be non-optimal. Perhaps you could evaluate the audible effect of that with a couple of tests. Tractrix, iirc, has the characteristic of terminating (at the large end) in a plane that is perpendicular to the axis of expansion. Thus, it would present essentially no discontinuity when mounted in an infinite baffle (such as a wall or very large speaker baffle).
As picowallspeaker implied, it would be interesting to include with your future efforts the study of the effectiveness of horns in front of various sorts of tweeter drivers.
It looks like 3D printing is the new frontier - what fun! Does your machine use spools of thin plastic rod as the consumable? Along the lines of re-filling ink or toner cartridges: is there a way to make your own by grinding/melting/extruding prior printouts or even recyclable household plastic items? I know this was (is?) common in the plastic injection-molding industry - even the less conscientious record manufacturers (or so I've heard).
And finally: thanks for the link to FreeCAD. Wow - from the screen shots, it looks like a very sophisticated program - and it runs on Linux too! I'll definitely have to check that out - maybe a defection from the TurboCAD camp is in my future!
Regards,
Wilf
Very nice effort - thanks for the picture and video.
May I suggest including a flange at the base - it would make it easier to balance it on your phone, plus it would assist sealing the joint with blu-tac or such, as mentioned. Sounds like you're on to bigger models - again, flanges at the mating joints should be helpful with alignment and connection of segments (whether the axis is straight or curved).
Sorry, I can't help you with the finer points of horn theory w.r.t. various profiles, cutoff frequency, etc. As I said though, the differences may not be hugely significant in your application - I'm sure that little speaker has a very limited low-frequency output, so I don't think any horn size or shape will drastically change that. But hey, have fun experimenting! Is the small end opening exactly matched to the speaker opening? Is the speaker sound emanating from a single opening, or is it an array of tiny holes? As I said, creating a step-like change in cross-section at that juncture would likely be non-optimal. Perhaps you could evaluate the audible effect of that with a couple of tests. Tractrix, iirc, has the characteristic of terminating (at the large end) in a plane that is perpendicular to the axis of expansion. Thus, it would present essentially no discontinuity when mounted in an infinite baffle (such as a wall or very large speaker baffle).
As picowallspeaker implied, it would be interesting to include with your future efforts the study of the effectiveness of horns in front of various sorts of tweeter drivers.
It looks like 3D printing is the new frontier - what fun! Does your machine use spools of thin plastic rod as the consumable? Along the lines of re-filling ink or toner cartridges: is there a way to make your own by grinding/melting/extruding prior printouts or even recyclable household plastic items? I know this was (is?) common in the plastic injection-molding industry - even the less conscientious record manufacturers
(or so I've heard).And finally: thanks for the link to FreeCAD. Wow - from the screen shots, it looks like a very sophisticated program - and it runs on Linux too! I'll definitely have to check that out - maybe a defection from the TurboCAD camp is in my future!
Regards,
Wilf
The phone case has 17 little holes for the speaker in a slightly oblong layout. The horn throat is 10mm across which just covers the longer side of this pattern. I'll try adding something to better seal the end. A flange is definitely a good idea too.
As for the input material for the printers, yes it uses a spool of round plastic filament. Currently the two most common print materials are PLA (polylactic acid)which is a plastic made from corn, and ABS (Acrylonitrile butadiene styrene) which is a very common plastic used in all kinds of commercial products, like LEGOs for example. The filament comes in 1.75 or 3mm diameter and various colors. I use 3mm diameter because it is cheaper per weight. It gets heated and extruded through a 0.5mm nozzle. Some people use smaller size nozzles for finer detail, but that also makes the prints take longer to complete. I get my filament from a site called ultimachine.com, but there are many other vendors selling similar product. When bought in 5lb quantities, it comes out to roughly $1 USD / oz. , including shipping costs.
The horn above was printed with a single wall thickness which ends up being about .65mm for my printer. The extruded plastic gets squished down a little bit, which is why it ends up wider than the nozzle.
As for recycling, there are a few people working on DIY style plastic recycling. The big important thing is to have a consistent diameter along the filament otherwise your prints will come out lumpy.
Here's one that looks promising MiniRecyclebot - YouTube
And there's also these guys Filabot: Plastic Filament Maker by Tyler McNaney — Kickstarter
As for the input material for the printers, yes it uses a spool of round plastic filament. Currently the two most common print materials are PLA (polylactic acid)which is a plastic made from corn, and ABS (Acrylonitrile butadiene styrene) which is a very common plastic used in all kinds of commercial products, like LEGOs for example. The filament comes in 1.75 or 3mm diameter and various colors. I use 3mm diameter because it is cheaper per weight. It gets heated and extruded through a 0.5mm nozzle. Some people use smaller size nozzles for finer detail, but that also makes the prints take longer to complete. I get my filament from a site called ultimachine.com, but there are many other vendors selling similar product. When bought in 5lb quantities, it comes out to roughly $1 USD / oz. , including shipping costs.
The horn above was printed with a single wall thickness which ends up being about .65mm for my printer. The extruded plastic gets squished down a little bit, which is why it ends up wider than the nozzle.
As for recycling, there are a few people working on DIY style plastic recycling. The big important thing is to have a consistent diameter along the filament otherwise your prints will come out lumpy.
Here's one that looks promising MiniRecyclebot - YouTube
And there's also these guys Filabot: Plastic Filament Maker by Tyler McNaney — Kickstarter
Well, i went ahead and published the script as is. I might add more horn profiles in the future as I learn more about them.
Parametric Acoustic Horns by thehans - Thingiverse
The sort of folded horn idea I have in my head will probably require a completly different approach to generating the model, so I'll just do that as a separate script if I ever get around to it. The idea is to fold the horn so that the wavepath travels in concentric rings. I would probably try to approximate a exponential expansion where the area would only change on each transition to the next "ring".
If I tried to make a continous curve... the math is too complex for me to think about right now.
I found at least one example which is kind of what I am thinking, only mine would do more folds, and I would keep the rings centered. http://www.diyaudio.com/forums/subwoofers/55377-folded-sonotube-tl-sub-complete.html
Except that is a transmission line design, and I would be trying to pretend mine is a horn, so I have no idea if that would work at all.
Parametric Acoustic Horns by thehans - Thingiverse
The sort of folded horn idea I have in my head will probably require a completly different approach to generating the model, so I'll just do that as a separate script if I ever get around to it. The idea is to fold the horn so that the wavepath travels in concentric rings. I would probably try to approximate a exponential expansion where the area would only change on each transition to the next "ring".
If I tried to make a continous curve... the math is too complex for me to think about right now.
I found at least one example which is kind of what I am thinking, only mine would do more folds, and I would keep the rings centered. http://www.diyaudio.com/forums/subwoofers/55377-folded-sonotube-tl-sub-complete.html
Except that is a transmission line design, and I would be trying to pretend mine is a horn, so I have no idea if that would work at all.
Yes! That's definitely what I had in mind ,but in a vague mode...Thx W_Osvald !!
Indeed , claiming the frequency/ driver diameter ratio without the power response
would be incomplete . For a cell phone speaker , which I think it may be very good for a tweeter , when listening very close , together with the shell of the phone ...as sound transmits even mechanically trough the vibrations ( as we can hear trough our in-bone constitution ),ah ! Ok ! Sorry , we were talking about music ,not speech .As I said , I think those are good tweeters ; I use two 3/4" 'button' loudspeakers ( the same that are inside notebooks and flat monitors)
with little horns / waveguides .
As your experiment demonstrated ,that when putting some surface near a sound source , some effects arise . The change from direct radiation ( which is not due to the holes in the plastic that put a 'cover' ) to horn expansion is clearly altering the media , changing directivity and amplifying in a certain manner . Speech reproduction may be a good test to make a decision for different horns ( also direct radiation ,,
) and distinguish from kinds of coloration, echo ,reverb that damage the sonic intelligibility .....
Sounds like you're visualizing something like a re-entrant horn, but with cylindrical sections? Re-entrant horns are commonly used in (voice) PA applications, and for car and home alarms. So, by keeping the area of each section constant, it might have some similarity to the horns described in this thread: http://www.diyaudio.com/forums/full-range/205729-spawn-olson-nagaoka-horns-next-generation.html
Interesting idea... Those Nagaoka (if that's the correct designation for them?) horns allegedly manage to extract amazing performance from relatively small drivers. Building a set of those for some Fostex FE103A on hand is another item on my endless project list.
Wilf
Nice work, I might try to reverse engineering to adjust the scrip to the tool I implemented in the FreeCAD too. This too might be from people interests here, see the website below and related thread. The major part of the design are for Subwoofers, but there are some for Top/Kickbin, so if I can find a way to implement curved profiles might be useful for some cases.
https://freeloudspeakerplan.rf.gd/
https://www.diyaudio.com/community/...plan-hornresp-integrated.391237/#post-7147619
https://freeloudspeakerplan.rf.gd/
https://www.diyaudio.com/community/...plan-hornresp-integrated.391237/#post-7147619