The first version of our 3D printed coaxial CX loudspeaker series was made using the Desktop Metal Forust method, which is, at the moment, too expensive for most DIY audio enthusiasts. Therefore, the CX2 was designed based on fused filament fabrication (FFF).
3D printing gives design freedom
The starting point for the design is simple but effective: sealed enclosure and coaxial driver. This inherently gives us controlled cone displacement in the low-frequency region and a coherent radiation source in the crossover region. 3D printing allows to easily implement two more acoustically beneficial geometries: large roundovers and compound curved walls. These translate into a Minimal Edge Diffraction Enclosure (MEDE™) and reduced panel vibrations, respectively. Curved walls mean that the loudspeaker requires a stand. This requirement can be turned into a benefit: the symmetric loudspeaker can be tilted or laid on its side when placed on a so-called Isopodd stand 3D printed from soft TPU material. 3D printing allows complex shapes at no extra manufacturing cost. For example, the front baffle is stiffened on the inside with a honeycomb strucure that acts also as support for the overhangs, but robs very little internal volume.
What you need to build your own CX2
3D files for 3D printing, sold on Etsy. Dimensions 210x273x170 mm, ~1200 grams of filament.
SB Acoustics SB13PFCR25-4 COAX or SEAS MP15 (contact us)
Two channels of amplification per speaker, ICEpower module recommended
Neutrik NL4MPR SpeakON connectors and fastON crimp connectors
4.2 mm wood screws
Bitumen or similar visco-elastic damping sheet and fibrous damping material such as pillow stuffing
Soldering capability (super easy)
How to build it
3D print the enclosures using the files mentioned above. It’s a single-piece print with no support needed. Wood-filled PLA or similar material is easy to sand and no other surface finish besides sanding is needed. If you are using the SEAS drivers you also need to print the TPU gasket/adapter. The speaker can be tilted and rotated when you print a small stand for it from TPU material. Some rubber feet on a plate will do the same job just fine. The assembly order is as follows:
Finish the outside of the enclosure and make sure the driver and SpeakON connector fit.
Line the inside with bitumen damping material and fill it quite densily with wadding.
Solder wires to the speaker drivers and crimp fastON connectors at the other end. Mark woofer and tweeter positive and negative wires.
Feed the wires through the SpeakON connector opening and mount the driver using wood screws. Use gasket/adaptor if you have the SEAS driver.
Connect the fastON connectors to you SpeakON connector and mount it.
Setup your bi-amping and crossover configuration. A good starting point for crossover frequency is 2 kHz. On-axis response will be bright, so keep that in mind when equalizing. Some toe-in may be beneficial.
3D printed loudspeakers do not have to be made out of plastic anymore. There is a new way of 3D printing wood called Forust. It is a binder jetting process where upcycled sawdust is used together with a binder to form the closest thing we have to 3D printed wood. We are particularly interested in the possibilities this offers loudspeaker manufacturers.
RD Physics CX1 – A coaxial loudspeaker
RD Physics has been developing speakers with full-range drivers for some time now and while they have their inherent benefits, it is time to look what coaxial drivers have to offer. The starting point was a spherical shape, which is known for its benefits. However, the limitations in build volume favored a shape closer to a rectangular cuboid. The shape of the CX1 has the largest possible roundovers, with the constraints imposed by driver size and maximum baffle dimensions. This is to reduce edge diffraction. The sides are compound curved to maximize stiffness. There is also internal ribbing to stiffen the enclosure without taking up internal volume like a sandwich structure would. The enclosure is made in two parts; the front baffle has a separate cover that conceals the driver flange and mounting screws. The driver is a proprietary SEAS unit designated MP15 (15 cm diameter). The idea is to have an external active crossover and bi-amp the loudspeaker via the Neutrik SpeakOn 4-pin connectors at the back.
3D printing a loudspeaker using Desktop Metal Forust method
The geometry files were sent to Forust for 3D printing. The chosen colour is “natural” with the artificial wood grain introduced during manufacturing. The result is a structure that looks like plywood. Parts can be ordered without the grain and with darker colours, too. The grain is more interesting, however, because various surface texture effects can be achieved by aligning the layers at low angles relative to the principal axes of the printed shape giving a zebra stripe effect.
Post-processing of 3D printed wood
The parts printed with the Forust method can be sanded smooth, but it is not like sanding natural wood. The surface can be varnished, but not stained. The Forust material does not absorb wood stain. It does not tolerate ethanol and perhaps other solvents either. Long-term exposure to water should be avoided, otherwise there will be is a sticky brown residue on the surface. Although a wooden look can be mimicked, post-processing is not similar to wood. Instead, it resembles the wood-filled polymers used in our previous builds. This is not a serious drawback, it just means that 3D printing skills are more useful than woodworking skills. In terms of aesthetics this is the closest thing available for increasing the acceptance of 3D printed loudspeakers in the audio community, where wood veneer is the go-to solution.
It is tempting to consider “room correction” with Digital Signal Processing (DSP) as a substitute for acoustic treatment. We implemented both in the same room to see what the effects actually are.
Experimental setup: DSP and acoustic panels
The setup used is a normal living room/home theater. The loudspeakers are Genelec 8351A active monitors with DSP and automatic calibration using a microphone and frequency sweeps.
The test setup with Genelec active monitors.
Five acoustic panels were placed in the room. They are mineral wool panels measuring 60x60x10 centimeters. Two of them were placed at the side walls in order to address first sidewall reflections and three of them were placed behind the listener by the back wall. The measurement point is also the normal listening point. Measurements were done with REW software:
The effect of acoustic panels and DSP on room response
Effect of acoustic panels (top) and DSP (bottom).
From the magnitude response we see that the acoustic panels bring down some of the peaks in the mid-range. When we then apply DSP and automatic calibration, we get attenuation of the low-frequency peaks caused by room modes. DSP does not really affect the mid-range and the highs. It only raises their level back to where it was earlier. Using DSP and equalizing for mids and highs would be very difficult, because notches and peaks are very narrow.
Waterfall charts show faster decay at mid-range frequencies when acoustic panels are applied.
Spectrograms show massive amounts of energy in the bass domain, where we have room modes affecting. Panels this size should not be very effective at long wavelengths according to the manufacturer and our magnitude plot. Yet, adding acoustic panels brings down the energy across the frequency range according to the spectrogram. DSP reduces the bass peaks which, of course, reduces the energy in that region. DSP brings up the mids and highs, so we can see slightly increased energy in that region, which leads to an evenly distributed energy across the spectrum of frequencies.
Spectrograms of the reference condition (top), with acoustic panels (middle), with acoustic panels and DSP (bottom).
Conclusions
So do you need both digital signal processing and acoustic treatment? Yes. Looking at the magnitude response, we see that DSP addresses the peaks in the bass region and adjusts for the overall level, while the acoustic panels address the mid-range frequencies. Looking at the energy spectrum, we can see that actually both acoustic panels and DSP even out the energy distrubtion across the frequency range. It is encouraging to see that placing only five panels has a measurable effect. Headphones are immune to room acoustics, but benefit from DSP. Check out our post on headphone DSP: https://rdphysics.com/2021/06/14/dsp-for-headphones
Video
This blog post can be found in video format as well.
Using a computer as your signal source gives you immense DSP possibilities. It does not cost a thing and reverting back is easy in case you do not like it. There is really no reason not to give it a try. A good place to start is here:
https://github.com/jaakkopasanen/AutoEq There you will find EQ presets for most headphones and links to applying equalization in your operating system using Equalizer APO. Some users may like the Peace add-on which can be found together with Equalizer APO. If you have issues with system-wide equalization, you may want to try a plug-in for your music player. We can recommend Foobar2000 (oldie but goldie) and Math Audio Headphone EQ.
Math Audio preset files for Porta Pro and HD800S headphones
Here’s the preset file to be used in the Math Audio plug-in when listening to Koss Porta Pro headphones. It’s based on Oratory1990’s EQ profile.
EQ for Koss Porta Pro based on Oratory1990’s measurements
We found the equalizing curve for Sennheiser HD800S to be too harsh, brightening up the sound too much. Therefore, the gains of the peaking EQ were halved and entered into the Math Audio plug-in. You can download it here:
We have got to remember that the EQ files found online are obtained using a measurement head and aiming at a flat frequency response. However, each individual has physically different ears and a flat frequency response may not be what we actually want. The recordings that we listen to vary and the type of music varies. The presets are a good starting point, but they should be tweaked to make sure the sound is to your liking. Are there any downsides to processing the signal? There could be some artifacts from filtering such as pre-echo and you could get added distortion from excessive bass boost, but as long as it sounds better to you then that’s all that matters. Give it a try!
The argument for headphones instead of loudspeaker as your main sound system is one that you don’t hear too often. Which is why we think it’s important to make it here. It all boils down to one root cause, and that root cause is the room. Let’s divide the consequences of the room into two categories: cost and sound.
Cost
First, speakers are played in a room you need more power. Power means power amplifiers. You need to buy expensive amps to power your loudspeakers. Second, you need to place those loudspeakers somewhere, so you need to buy stands. Or if they are floor-standing speakers you need to buy feet. You need to connect them with cables and buy other accessories. Third, you need to acoustically treat your room, so you need to buy acoustic panels, diffusers, bass traps etc. Fourth, you need to buy presents to your spouse because you’re placing the speakers in the middle of the room.
Sound
You can buy good loudspeakers and ruin them by placing them in a bad listening environment. Optimally, you would have the loudspeakers and the listening position at least two meters away from the nearest wall. However, that is seldom even possible in the available space. You would need a large room. And with this kind of placement, a livingroom quickly becomes a listening room only. Headphones, on the other hand, have multiple benefits compared to loudspeakers:
Single point source
No crossovers
No sweet spot or particular listening position
No room effects
Tonal balance can be fixed using only DSP
Some of the drawbacks often stated include poor sound stage or imaging. People say that it sounds like the sound is coming from inside one’s head and it doesn’t feel like you’re at a concert. It is a matter of personal preference, but we suggest looking at headphone listening as something separate and different from live events or loudspeaker listening. It is our subjective opinion that crossfeed will not correct for this phenomena and only makes the sound worse. Another common argument is that there’s no physical sensation of bass. While that is true, the pros outweigh the cons.
Recommended hardware
Which ever headphones you use, applying equalizing with the help of DSP is definitely worth considering. Check out our post on the topic:
Our previous 3d printed subwoofer, the SW1, is a 13 liter subwoofer with a 6.5″ driver, a matching passive radiator and a plate amp. We wanted to develop something smaller that would still offer the bass extension that satellite speakers so badly need. The result is the SW2 using a Tang Band W5-1138 5″ long-throw driver and the same Dayton Audio DSA175 passive radiator as in the SW1. The enclosure is now only 5 liters and much easier to fit on a desktop. The passive resonator allows tuning the resonance frequency to avoid overlap with room modes, for example. The spherical shape is optimal for material use and stiffness. Combined with the small diameter driver with large surrounds, the appearance is quite unique. If a traditional box is what you want, then this build is not for you.
Measurements
The measured resonance frequency of the passive radiator indicates that some air-coupling occurs due to the downward firing placement. Simulated resonance frequency matches the measured value (53 Hz) when 16 grams of added mass is used. Mass can be further added using washers to tune the response. In practice, the frequency response starts to drop below 50 Hz. The Arylic amplifier offers DSP capabilites and using a computer as the source allows unlimited DSP with zero cost. Therefore, frequency response in not that meaningful especially when considering the room effects, but we have included some measurements to give an idea of the natural response especially around the lower cut-off.
3D printing
The enclosure is printed in one part (234 mm diameter) and takes approximately 1.5 kg of filament. Print time is about 48 hours. The mass can be increased by lining the walls with sound deadening mat. Although the external wall is spherical, there is a cylindrical inner wall that braces the woofer to the passive resonator and, thanks to a single curvature surface, allows easy installment of thick sound deadening mat. The drivers are fastened using 4.2 mm wood screws. There is a geometry file for a gasket for the woofer which can be printed from TPU. Traditional gasketing methods will work, but the 3D printed gasket is seamless and has the screw hole pattern accurately incorporated. The binding posts are recessed deep into the enclosure and only accept banana plugs in that configuration. An O-ring under the binding post washer is recommended and there is a chamfer for it. 3D printing using a wood-filled filament allows easy sanding for a smooth surface finish. The photos show 15 minutes worth of post-processing making this a very easy and fast build without compromising in function and looks.
Images
Sound quality
The subwoofer was compared to the much larger, THX certified Logitech Z623 subwoofer. The sound is very similar, but in a much smaller package. The SW2 is a great companion for small satellite speakers and brings fullness to the bass. Electronic music will benefit from the “boom” offered by this small unit, while other types of music may require turning down the level a bit for a tighter bass.
Our 3D printed full-range speakers needed something to beef up the lower end of the frequency spectrum. We set out to design a compact subwoofer that can be used together with our FR3 speakers. The result is a 13 liter enclosure with a 6.5″ driver, a matching passive resonator and a plate amp. The passive resonator allows tuning the resonance frequency to match room modes, for example. The plate amp can power satellite speakers and has a fixed high-pass filter. The low-pass cut-off frequency for the subwoofer can be adjusted and the level too, which means that this system can be easily mated with signal sources that do not have equalizing or DSP capabilities in themselves.
3D printing
The enclosure consists of two parts, which are glued together after printing. Total print time is about 100 hours and uses about 4 kg of filament. Support is only needed for the small recess where the plate amp is mounted. Dual-material printing is not needed. The mass of the enclosure can be increased by filling the walls with epoxy through the holes in the back. A geometry file for 3D printing a matching funnel is provided, too. 2 kg additional mass can be obtained this way.
The 3D files can be found on Thingiverse for free:
It is striking how often 3D printed speakers take the shape of a sphere and that is also how RD Physics started with the FR1 full-range speaker. What are the benefits of spherical loudspeaker enclosures and why are they so popular?
Rigid and void of panel vibrations
Minimum material use for a given volume
Potentially avoid edge diffraction
Omnidirectional up to a relatively high frequency and controlled baffle “step”
Aesthetically pleasing with a single circular driver
Difficult to manufacture any other way than 3D printing
As RD Physics has extensive experience in these types of enclosures, we have decided to share our learning in one blog post. The models are presented in chronological order allowing the reader to understand the development that took place over the years of building and listening to various versions of the FR. Most of the designs are offered open-source to the community.
FR1 – Full-range bliss in a compact form
The FR1 is a spherical (180 mm diameter) full-range loudspeaker with a sealed enclosure. Internal wall stiffeners are used in order to maximize internal volume as opposed to simply increasing wall thickness, which robs internal volume. We use Noise Killer paint to both seal the enclosure and also to add mass and damping. The sound of the FR1 speakers is very unique and quite tricky to get the most out of. The full-range emitters are very directional and the listening distance also changes the sound markedly.
Model
RD Physics FR1
Driver
Mark Audio Alpair 6M 2.5″
Enclosure
3 liters sealed
Material
UPM Formi3D
Construction
Internal webbing with Noise Killer damping
Tilt
Fixed at 15 degrees
FR2 – Exotic carbon fiber skin reduces resonances
Additive manufacturing (AM) has many benefits over traditional construction methods, such as design freedom, fast product development, and integration of functions into one part. There are drawbacks as well. The plastic AM parts tend to be low in mass and not very stiff. Air-tight walls are sometimes difficult to achieve, too. Adding mass by increasing the fill density of the print is not a good solution, since it adds build-time and material cost. Stiffeners and bitumen paint were used in FR1. However, the stiffeners were cumbersome to paint with bitumen and it did not add significant weight. For FR2, we used the vent as a part of the mechanical structure and used a thicker wall. But some additional means were needed to bring 3D printed enclosures on par with traditional cabinet materials.
Carbon fiber in loudspeaker building
Dry carbon fiber tow was wound around the enclosure and then wetted with epoxy resin. The composite skin was sanded after curing and additional coats of epoxy were added. The result is a unique unidirectional carbon fiber surface finish. The composite shell adds mass and stiffness to the enclosure. The loudspeaker sits on four feet printed from TPU material, which allows rotation.
Model
RD Physics FR2
Driver
Mark Audio Alpair 7MS 3″
Enclosure
5 liters vented
Material
UPM Formi3D + CFRP
Construction
Structural port and carbon fiber skin
Tilt
TPU feet allow tilt ~0-15 degrees
FR3 – Metal-filled filament for mass and rigidity
The FR1 used internal ribbing and Noise Killer paint to reduce enclosure resonance. The FR2 used an external carbon fiber shell. Both approaches were a bit cumbersome and laborious. For the third version, we wanted to fully use the capabilities of 3D-printing. Therefore, a high-density metal-filled filament was used an internal gyroid-shaped support was used even where overhanging surfaces would not have required it. In addition, height and tilt can be adjusted using three threaded rods that form a tripod. The finished enclosure with three 14 mm trapezoid-threaded nuts bonded to it weighs 1.2 kg. The RS100 drivers have a distinct on-axis peak at the upper treble, which actually works nicely for those who like a bright sound. Those who don’t should toe-in the speakers a bit.
Model
RD Physics FR3
Driver
Dayton Audio RS100 4″
Enclosure
2 liters sealed
Material
Colorfabb steel fill
Construction
Gyroid infill as an internal stiffener
Tilt
14 mm leadscrew tripod
FR4 – Refining the concept further
The metal-filled filament used in the FR3 was too brittle and difficult to post-process. The FR4 uses wood-filled filament, which is more ductile and easier to sand if needed. The surface is quite nice straight out of the printer thanks to the matte surface. A quick touch with an orbital sander gives a smooth finish. Leadscrew nuts are bonded into recesses in the enclosure and allow for adjustment of the legs. Small TPU feet can be printed and placed at the ends of the leadscrews in order to avoid scratching the desktop. These are satellite speakers and need a subwoofer to complement the lower frequency spectrum.
Model
RD Physics FR4
Driver
Tangband W3-1878 3″
Enclosure
1 liter sealed
Material
AddNorth Textura
Construction
Gyroid infill as an internal stiffener
Tilt
12 mm leadscrew tripod
FR5 – Returning to square one
Reviewing the FR project so far, we came to the conclusion that all things considered, the original FR1 is the DIY project that was the most fun to build and listen to. It’s simple but rewarding once dialed in. For the FR5 we went back to basics by ditching the tripod and returning to a simple white spherical enclosure. The tilt adjustment is handled by a TPU mounting ring that allows a large adjustment range. The Scan Speak 10F driver is one of the best for voice reproduction, but our subjective view is that it needs a tweeter in addition to a subwoofer making it suitable for three-way builds only.
Model
RD Physics FR5
Driver
Scan Speak 10F 3″
Enclosure
2 liters sealed
Material
AddNorth Textura
Construction
Stiffeners and alu-butyl sound-deadening mat
Tilt
TPU ring +-30 degrees
3D files and components
In the table below you’ll find links to the drivers used in each version as well as the geometry files needed for slicing the toolpaths. Support us by clicking on the Soundimports affiliate links before buying anything from them (we get a small commission and it won’t cost you a dime). Thank you!
We wish to thank UPM for the Formi3D materials and support. Photos taken by J-P Virtanen and Markus Markkanen. Erell Bodinier handled the carbon fiber skinning.
