I would endure poor people skills to get the right competent people. I taught speaker design for one semester, the head of the department stressed the importance of people skills to the students. It was my view that you did not even come into class if you can get all assignments done well, and the scoring system I laid out was very precise. They found out I meant business when almost half the class flunked. And they flunked simply because I could not give a score for nothing turned in, and that was how it averaged out. Did not expect to babysit graduate students.
I have a question that does not require people skills: how does a section of golden spiral perform? Pretty much like any exponential, or no?
Some Approximations
To a first approximation, would be the Tractrix.
Perhaps elements of JBL Biradial Horn as well.
The Le Cleac'h horn, approximates Euler's Spiral.
Here curvature is a function of curve length.
Regards,
WHG
To a first approximation, would be the Tractrix.
Perhaps elements of JBL Biradial Horn as well.
The Le Cleac'h horn, approximates Euler's Spiral.
Here curvature is a function of curve length.
Regards,
WHG
Has anyone looked into 3D printing to do more experiments. I met someone who did a demo and inquired about a larger version. Thinking about getting a 1 cubic ft volume capability as a minimum just for continued experiments.
1 cubic is enough for some things 100%, but 3d printing the throat and mounting point assembly up to a 12" depth, would be very handy even for larger horn constructions. Depending on the flare, you might even add bolt-together flanges and do a second section, for something with a long throat like a lecleach.
From what I've seen, 3D printers big enough and reliable enough to do things that large will run you over $30,000.
Much more than that. The price jumps drastically going from consumer-grade to industrial-grade. The largest consumer unit right now is probably the CubeX.
I talked to some place that use Stratasys machines, they seem to do 0.01mm layer step, but sending a horn to do be printed like that was just as expensive as I had mine done using the traditional method, and was still smaller than 1 cubic ft. Talked with someone demoing a few DIY units that delivered what they called 0.05 mm step layer, seemed sufficient for studies for measurement purposes. Talked with them about getting it up to 1 cubic foot size, it seemed doable. Still surveying what other options are out there.
how about stacking CNC routed birch plywood rings? there are lots of CNC wood routers out there now, hobby desktop CNC routers can be bought for the cost of a larger waveguide's 3d printing material charge
and super accuracy, resolution isn't obviously needed with 1/16" already being ~ 1/10 wave @ 20 kHz
and super accuracy, resolution isn't obviously needed with 1/16" already being ~ 1/10 wave @ 20 kHz
Last edited:
For those really large horns, wood NC is fine. For more frequent testing, when you think about the setup work, cleanup work, waste disposal, it is really more labor intensive, and the slope of the horn make it inaccessible for stable machining. If you also build a compression section into the horn, then you need to flip it over. Really not suitable for research stage work.
I've been poking my head into this thread since it started but have lost track. When you guys are talking about making horns, are they round? Like something that can be made on a lathe?
waveguide dimensions
I am thinking about adding a waveguide to a little tweeter.
It operates from 6kHz upwards.
The start of the wave guide has a radius, EG suggested around 2" (50mm) for this first transition to the conical "horn".
The mouth then has a a bigger radius to transition from the conical "horn" to the speaker baffle, EG suggested 4" (100mm).
But what about the actual conical "horn", the bit that gives the beam shape/width.
How long does the "cone" part need to be?
What range of half angle of cone is acceptable?
Does the angle and/or length depend on the range of frequencies that the driver must handle?
Am I correct in thinking that the throat radius and the mouth radius are there to attenuate/eliminate diffraction at a change in surface angle?
I am thinking about adding a waveguide to a little tweeter.
It operates from 6kHz upwards.
The start of the wave guide has a radius, EG suggested around 2" (50mm) for this first transition to the conical "horn".
The mouth then has a a bigger radius to transition from the conical "horn" to the speaker baffle, EG suggested 4" (100mm).
But what about the actual conical "horn", the bit that gives the beam shape/width.
How long does the "cone" part need to be?
What range of half angle of cone is acceptable?
Does the angle and/or length depend on the range of frequencies that the driver must handle?
Am I correct in thinking that the throat radius and the mouth radius are there to attenuate/eliminate diffraction at a change in surface angle?
I've had a fair few prototypes created for work lately for some of our development products.
We've opted to go with CNC'ing as it gives a far greater finish than 3D printing.
The steps on a 3D printed part are noticable (even on a ~$90K ProJet printer).
With some of the materials, you can dip them in acetone to smooth them over.
I've been contemplating having some waveguides printed, and they would most likely need to be done in 2 parts (ie: cut your waveguide in half) and then adhered together.
It certainly would be a quick option, but having a prototype made of a part ~20cm x 25cm x 5cm currently costs us about ~$400 AUD... makes for an expensive waveguide.
We've opted to go with CNC'ing as it gives a far greater finish than 3D printing.
The steps on a 3D printed part are noticable (even on a ~$90K ProJet printer).
With some of the materials, you can dip them in acetone to smooth them over.
I've been contemplating having some waveguides printed, and they would most likely need to be done in 2 parts (ie: cut your waveguide in half) and then adhered together.
It certainly would be a quick option, but having a prototype made of a part ~20cm x 25cm x 5cm currently costs us about ~$400 AUD... makes for an expensive waveguide.
What if you had the negative space inside the waveguide CNC'd inside a big block of rigid foam, and then used the foam block for testing, instead of cutting the positive forms and making a mold?
The reason I like 3D printing concept is that it seems pretty much print and forget until it gets done. Not really smooth enough for real world listening, but sufficient for experimenting with different design concepts for technical evaluation.
The surface finish is not too bad with today's models. You might be able to get up to 20 kHz and each step would only be a 1/8wl discontinuity (math from memory).
Some light sanding should make it even smoother. But not really ideal for sound reproduction. Generally glossy finish should be the best. At least I had seen someone post experience on the web in such conclusion.
The angle you choose will fit the application but 90 degrees makes sense as a baseline in a rectangular room, and maybe to 60, where the horn can get long covering the same band.
If the straight section were extended to cover the near-field of the low end (something I use as a starting point), you might find in a higher frequency waveguide that the mouth radius starts fairly close to where the throat roundover ends.
The mouth radius alleviates the reflection due to the impedance mismatch between the mouth and the room. At higher frequencies this is more localised toward the edges. The throat roundover is more of an attempt to follow the waves with the walls..if the driver were producing a spherical wave (as if from a point source) it would fit into a conical horn without the roundover at the throat.
Last edited: