About 17 years ago, I embarked on an ambitious loudspeaker project — ambitious for someone just starting out — a pair of two-way bookshelf speakers and a subwoofer. Those speakers are still being used today, and I thought to myself that enough time has passed for me to be able to draw some solid conclusions.
A minor detour
It was 2003-ish. I’d already had already completed two simpler speaker projects, and I say that with a bit of nostalgia. At the time, I was quick to dismiss my speakers as being awful: an 8″ poly-cone woofer in a ~30L sealed box made of 18mm plywood, and 25mm fabric dome tweeter for the highs.
Looking back though, they only had a few minor flaws, but I was eager to get rid of them. The drivers were low-cost generics from an electronics store that was popular/existed (Dick Smith). I don’t think the woofer’s magnet had a vented pole piece (the 10″ version did), which seemed to limit the bass output somewhat, and the tweeter was a bit screechy.
Solid materials
I was a little hung up about so-called cone breakup, and whatever else might’ve been distorting the sound, so, for the big project, I went all-out and bought some 7″ Accuton woofers and 32mm tweeters. The super-stiff ceramic cones caught my attention, as I was certain that those pesky resonances had to be be pushed up in frequency, well out of the pass band. Similarly, everyone (“everyone”) in online fora seemed to be in a race to fix unwanted panel resonances by taking inspiration from submarine and nuclear reactor designs.
I hopped on board that band-wagon and got some exotic hardwood with the trade name ‘bubinga’, which was used for building the necks of bass guitars because of its extreme stiffness. It behaved more like a block of aluminium titanium than wood, and the circular saw, which I was already terrified of, kept screaming as though it was cutting through nails.
The resulting 15L boxes were sleek, rustically beautiful, and much stronger than they needed to be. Early on, a couple of glue lines broke when there were large changes in humidity. IIRC the glue was letting go of the wood, allowing it to bow outwards while still connected along one or two edges. It may have been either epoxy or acetone-based craft glue that won the strength battle, and I have an odd suspicion it may have been the latter. Epoxy gets a lot of praise, but in this case it may have been too brittle, and the wood too oily. What was surprising to me was that the change in sound quality was mostly just a change in the tonal balance. I thought it would be worse!
Damping was a battle on two fronts: taming the air resonances inside the box, and preventing the wood itself from vibrating. I experimented with lining the inside with heavy carpet, poly fill, and blocks of styrofoam. They were all moderately satisfying, but never quite fixing what I wanted them to fix.
A 12″ subwoofer in a hefty 60L sealed box of 25mm plywood provided bass.
After learning about active filters at school, I developed some op-amp based filters: a 2.5kHz crossover frequency, a 5kHz notch filter for the mid-woofers, a simple shelving filter for baffle-step correction, and a 7th order Bessel filter at 120Hz for the sub. Despite the ‘high’ cut-off frequency, the sub was superbly muffled and lacked the the mid-bass boom that seemed standard with contemporary 2nd/4th order Butterworth filters on plate amps.
My design philosophy at the time was that the purpose of the box was to ensure a pure monopole output, by absorbing the unwanted back-wave with modest padding.
If I could do it all again…
- Plywood was a good choice.
Wood wasn’t necessarily a bad material, just a bit more resonant than plywood. Dipole resonances were also an unresolved issue. One solution could be to arrange pairs of speakers in a ‘boxer’ configuration to cancel out some of the vibrations. Soft adhesives like acetone based craft glue could also help to limit conduction from one panel to another.
- Don’t bolt to the front.
With most ‘box’ type designs, there seem to be some conflicting requirements, and the conflict often remains unresolved. The air inside the box is pressurised by the speaker, so the walls should of course be strong enough to not flex under the very tiny pressure. Another challenge is that the motor will vibrate during normal operation, and someone (possibly a committee) decided to standardise the mounting point so that, by default, the vibrations are conducted with maximum efficiency to the outside bulk of the box, and then to the surrounding air.
I’ve heard arguments that there are better materials out there, which are more absorbent, so if there’s an issue with vibration, just add more MDF, or that additional bracing is the cure. That approach seems like an attempt to reduce one problem by adding another — more mass. It takes the issue of box vibration as inevitable. Embarrassingly for me, it has taken a while for me to appreciate and do something about this issue. I’d always kind-of known, and made sure to add rubber gaskets to decouple that interface between metal speaker basket and the wooden baffle, hoping that it would also absorb ‘some’ of the vibrations. But I never got round to improving the experiment: suspending the magnet inside the box, and gently balancing the speaker on a gasket to minimise the forces acting on the baffle in the first place.
- The boxes would be bigger and wider, just like the 30L ones.
The sleek, slimline 15L boxes had a correspondingly ‘thin’ sound, which I augmented with an active bass-boost shelf of about 6 dB below around 700 Hz. Larger boxes would have higher sensitivity, just by virtue of how their outer dimensions help to displace the surrounding air.
- Just say ‘no’ to mono subs.
A 12″ mono sub, tucked “out of the way”, was a terrible idea. It was a 1000 times better than no sub, but it really should’ve been stereo. Although I was pleased with how well it seemed to blend in with the other speakers and disappear, there was no standard for how stereo recordings ought to be converted to mono at low frequencies. Should I use one channel and double the amplitude? Or maybe double the power? Or maybe add the two channels together, so that differential signals would cancel out (which is certainly possible, for instance with big-band bass drums)? These were unanswered questions, probably best left to the experts who gave us “home theatre”.
A later project uses a stereo pair of 10″ woofers for bass, and it has much better integration. The project itself is in an unfinished state at the moment, with the woofers crossed at 1kHz because the 3.5″ wide range speakers were a bit rough below that point. The 40L boxes are as yet without internal bracing because they were only designed to be used up to about 200Hz in a 3-way system. The modification that I have in mind seems risky because it would involve making the cutout bigger, and balancing the edge of the basket on an air-tight gasket. There would be no going back to using bolts.
- Dipole or aperiodic vents
The Accutons always seemed to reveal a slight harshness — the damping in the box was inadequate, but the design wasn’t physically big enough for some of the absorption techniques that came to mind, like anechoic wedges, or multiple chambers with resistive vents. The back-wave was screaming to be “let out”. never seemed to don’t seem to like sealed boxes.
Conclusion
Good speakers, but… there a loads of things that I’ve learned, and alternative ideas that come to mind. It’s a long post already, and I’ve already delayed it by adding more, so I’ll cut it short here, and may just do some follow-ups later. Remind me to add some photos.