The real challenge of this build comes from trying to turn a 3D CAD model into a physical object. The benefits of the CAD modelling are huge as they largely remove the need to do math for some of the complex angles in a 3 dimensional space however they do present another problem: how do you accurately make a 46.8° cut? My answer: CNC.
I found a CNC company in Melbourne who would do small batch machining. There were two major challenges with this machining: accurate angles and having to cut easy angles and ones where we need to 'undercut' the piece using a 3 dimension (not 4 dimension) CNC.
The first problem was solved by using an extremely small bit and taking up to 64 steps per angle:
To address the second problem the company designed a jig which would allow them to turn over the piece after the initial cut to allow another stepped cut from the top.
It is so nice to be able to see your 3D CAD model become a physical object so quickly, easily and accurately. I can only imagine how incredible the next revolution of additive manufacturing via 3D printing will be.
Once the first step was done I delivered the pieces to the next manufacturer who previously helped me build my ea loudspeakers: Aranmar Acoustics. Given the accuracy of the CNC everything (mostly) just worked.
The first test fit - and a small sigh of relief.
There were still a few problems to solve: the first on how to mount the drivers to a flat face without interfering with the driver as I had seen some photos from another user of the Celestion midwoofer that ruined the surround due to friction against a flat surface. The anwer: wooden gaskets.
The second problem was to allow better airflow for the midrange drivers by routing out some of the panel in the midrange chamber. Once again: go CNC.
If you look closely at the midrange gaskets you can see they they are not completely round. This was done because a CNC model fits together so well that it is easy to forget that you need a way to mount the drivers. Luckily, with a bit of tweaking, the guys worked out a way to get everything to fit. Even then, everything is so tight.
During this time the guys had spraypainted them. It was not until everything was screwed in and drivers mounted that we really new that everything would fit. A major relief. Look how close the compression driver mounting bolt fits with the midrange gasket:
Finally it was time to mount the drivers.
Finally it was assembly time.
You can see that along the top and bottom edges there are many bolts. These allow the top and bottom panels to be removed to allow me to get into the cabinet for finishing and to wire everything up. Because the panels were extended the side of the cabinets they are effectively internal bracing whilst allowing the 10" drivers to fit. The Danley Sound model does really well to get such big woofers into his designs.
After almost 10 months in the shop (these were a low-priority build) they were loaded up into the car and taken home. If they were wired up I probably would have had one of the loudest cars in the Southern Hemisphere.
Where I may have made a mistake was designing them a little bit too big. A 700mm x 700mm box may not look so big on a CAD model but is quite intimidating when it is facing at you. Even though I tried to get them to increase the SAF by painting a 'clean' colour it seems to have failed.
Unfortunately when I got them home I noticed one of the horn joins had not glued correctly so I was faced with the prospect of trying to get glue down into a ~1mm gap. After a few awkward conversations with some pharmacists I had an answer:
If I had more clamps I could have done the whole lot in one go but spacing them out seems to have done the job and this super-strong glue should hold no worries.
The next job was to try to increase the mass of some of the internal surfaces by lining them with a liquid Bitumen Rubber product. When mixed with small stones you get a nice uneven surface with plenty of added mass and the benefit of increased water-proofing (a.k.a. sealing). In the photo above you can also see the start of my efforts to improve sealing by making a gasket with closed-cell foam which will be sandwiched between the top and bottom covers when bolted back on. This product is intended to stop drafts for leaky windows but serves this purpose very well.
I also used some car audio self-adhesive dampening foam/rubber on the drivers to hopefully prevent any ringing in the thin metal basket. The leftovers were stuck to the walls particularly where the ports were facing the walls.
The final step was to do the wiring. These are wired using 8 pole Neutrik connectors (these are beautiful hardware) and wired with:
Tweeter connected with 30μF in series as a protection from DC thumps which occasionally occur when using active crossovers. This will create a first order highpass at approximately 700Hz which will be below the crossover frequency.
Midrange drivers connected in a series-parallel setup (i.e. two sets of two drivers in series connected in parallel) to produce an approximate 5.8Ω load.
Woofers connected to their own connectors then wired in parallel at the amplifier. These are 5.6Ω nominal drivers so it presents a 2.8Ω load to the amplifier. By allowing the connections to be determined at the amplifier it is possible to wire these drivers to two separate channels on the amplifier which would both see 5.6Ω each. My plan is to eventually switch to a Hypex style class-D amplifier which would have no problem driving 2.8Ω.