The goal of this project is to design high-fidelity speakers that extend from 100 Hz to 20 KHz using only sound exciters as drivers. You may have heard of sound exciters, which are devices used to excite various materials and turn them into speakers. One good model of such a transducer is made by Dayton Audio, licensed through NXT technologies and distributed at Parts Express. This transducer (the DAEX25) will be the driver used in this project.Ideally, sound exciters are coupled to something that is:
- Larger, for better bass response
- Not dense, for higher efficiency
- Rigid (won't bend easily)
Most projects I have seen use materials such as foamboard, Gatorfoam, MDF, or even cardboard. Generally, these materials seem to satisfy the requirements I just listed. Using 20" x 30" foamboard panels from Michael's or Walmart indeed do work, producing sound with very high output power and good bass response.
But in all my trials, I was never satisfied with the results using any foam-based or particle composite material. Even after attempting to balance their response with equalization, they had consistently poor transient response, and were incapable of producing much above 8 KHz.
The next material I tested was an acrylic panel, 0.093" thick with dimensions of 18" x 24" (purchased from the Home Depot). In this sizing, the panel is reasonably stiff and not excessively heavy. Immediately after coupling the transducer to the panel, I noticed exceptional clarity in the sound. There was no muddy character like that in the foam panels, and there were definitely highs to be heard. But as it was, the sound at best reached to 10-12 KHz.
A notable drawback to acrylic is its greatly reduced efficiency, and difficulty with low bass. It must be driven with much higher power to achieve reasonable output levels compared to foamboard.
To improve the high frequency response, I attached a rectangular sheet of aluminum foil around the driver, on the outward (front) side. This rounded out the response on the high-end, and did not compromise the wide dispersion pattern that the panel already had. With two drivers mounted to the same panel, the foil "tweeter" achieved the same improvement still. Note that foil-tweeting does not work if you are using a soft material like foam for your panel. I have only found it to work well on acrylic. Also, the two drivers are currently wired in series, giving a relatively high impedance of 16 ohms. If higher power and lower impedance is desired, wire up another series-pair in parallel with the current one.
As I was very satisfied with the sound quality (rivaling most high-end systems I have sampled), the next step was to simply build a nice stand for it. What I came up with is shown below. The frame is built from 3/4" PVC joints and sections, including one T, one 90-elbow, a two-output 90-elbow (at the base), one straight coupler, and two 45-degree elbows at the ends of the feet to tilt the stand upward. The main pole is strengthened by putting a 3/4" wooden dowel through the PVC tubing, and the panel is coupled to the stand by hot-gluing the T-joint to the top of it.
The stand, after being assembled, was painted a glossy black to obscure the less-than-amazing PVC manufacturing labels. A small aluminum square was placed over the T-joint coupling spot to cover the hot-glue that would otherwise be visible through the acrylic.
I have assembled a stereo set of these speakers and tested them with some of my favorites (Spock's Beard, Porcupine Tree, Transatlantic, and Frost) and I am in the process of collecting more useful acoustic response data so that I can describe the speaker's performance with more than glowing adjectives.
Suffice to say, acrylic produces the best sound quality of any material I have tested with sound exciters, at the expense of considerable efficiency. However, with two drivers per speaker, they can nevertheless produce sufficiently loud sound for small-to-medium sized rooms, and display excellent imaging and wide dispersion.
Some final recommendations/notes:
This material is copyrighted by Jeff Chiles, 2011. Please contact me with any IP-related questions, or if you're interested in having me build you a set.
But in all my trials, I was never satisfied with the results using any foam-based or particle composite material. Even after attempting to balance their response with equalization, they had consistently poor transient response, and were incapable of producing much above 8 KHz.
The next material I tested was an acrylic panel, 0.093" thick with dimensions of 18" x 24" (purchased from the Home Depot). In this sizing, the panel is reasonably stiff and not excessively heavy. Immediately after coupling the transducer to the panel, I noticed exceptional clarity in the sound. There was no muddy character like that in the foam panels, and there were definitely highs to be heard. But as it was, the sound at best reached to 10-12 KHz.
A notable drawback to acrylic is its greatly reduced efficiency, and difficulty with low bass. It must be driven with much higher power to achieve reasonable output levels compared to foamboard.
To improve the high frequency response, I attached a rectangular sheet of aluminum foil around the driver, on the outward (front) side. This rounded out the response on the high-end, and did not compromise the wide dispersion pattern that the panel already had. With two drivers mounted to the same panel, the foil "tweeter" achieved the same improvement still. Note that foil-tweeting does not work if you are using a soft material like foam for your panel. I have only found it to work well on acrylic. Also, the two drivers are currently wired in series, giving a relatively high impedance of 16 ohms. If higher power and lower impedance is desired, wire up another series-pair in parallel with the current one.
As I was very satisfied with the sound quality (rivaling most high-end systems I have sampled), the next step was to simply build a nice stand for it. What I came up with is shown below. The frame is built from 3/4" PVC joints and sections, including one T, one 90-elbow, a two-output 90-elbow (at the base), one straight coupler, and two 45-degree elbows at the ends of the feet to tilt the stand upward. The main pole is strengthened by putting a 3/4" wooden dowel through the PVC tubing, and the panel is coupled to the stand by hot-gluing the T-joint to the top of it.
The stand, after being assembled, was painted a glossy black to obscure the less-than-amazing PVC manufacturing labels. A small aluminum square was placed over the T-joint coupling spot to cover the hot-glue that would otherwise be visible through the acrylic.I have assembled a stereo set of these speakers and tested them with some of my favorites (Spock's Beard, Porcupine Tree, Transatlantic, and Frost) and I am in the process of collecting more useful acoustic response data so that I can describe the speaker's performance with more than glowing adjectives.
Suffice to say, acrylic produces the best sound quality of any material I have tested with sound exciters, at the expense of considerable efficiency. However, with two drivers per speaker, they can nevertheless produce sufficiently loud sound for small-to-medium sized rooms, and display excellent imaging and wide dispersion.
Some final recommendations/notes:
- Keep the panel a foot or two from the nearest wall to prevent cancellation of low frequency sound.
- To extend bass response, add a low-end boost on the digital EQ and use a subwoofer.
- If you're building it yourself, expect to pay around $35 per speaker.
- Break them in slowly, and gradually increase volume to determine where their limits are.
This material is copyrighted by Jeff Chiles, 2011. Please contact me with any IP-related questions, or if you're interested in having me build you a set.
Great stuff, Jeff! How does this compare, cost-wise, to commercially available sound systems?
ReplyDeleteHey Phil!
ReplyDeleteNXT makes patented flat-panel speakers using similar drivers that sell for a few hundred dollars each. Those speakers work on somewhat different principles from the ones I have designed, though.
To compare sound quality, between these and non-flat-panel speakers, I need to do some more tests. However, I can say they have some unusual benefits such as dipole response (meaning sound comes out the back and front equally), wide dispersion (the sound covers a broader volume) and good transient response (detail in plucked instruments, vocals, drums).
Once I have the numbers, I can say more definitively how they compare in value. For now, I am extremely satisfied with the small investment I made.
Cool project idea. It's inspiring me to concoct my own experiments!
ReplyDeleteTwo questions about it though:
What made you decide not to use H-Frames? (price too high, or too much work, or both I would assume?)
Also, how much did you spend on the testing, materials, etc...? Or how much did the finished product cost?
Hey max,
ReplyDelete1) Do you mean like an H-frame subwoofer cabinet w/ OB full range? Or do you mean an H-frame for suspending the panel? Generally you want as few mounting points as possible on the stand for the panel, so it was better to have it coupled to one point with the Y-stand I have now.
2) Blog's updated with this info now. Although testing costs are hard to say...this design went through a number of stages and there was a good amount of wasted material.
Haha fair enough. I meant suspending the panel. I look forward to your future schemes.
ReplyDeleteI wonder how 4 of these would sound on a motorcycle windshield?
ReplyDelete