Before I start remembering what has been going on since the last blog update on my soundbar (which was 7 months ago), I wanted to explain why I am keeping the outdated design that is present in the first blog. I do consider these blogs a journey that accurately demonstrate my learning process and advancement in this project that, technically, crosses multiple disciplines of engineering. A lot of them were learned in school, such as DSP, Transducer's Theory, and microprocessors. However, there are many that I had to spend after school hours till 3 a.m., going through discussions on Stack Overflow, Autodesk forum, and countless Youtube videos to be able to work out myself. For example, CAD modeling, using Fusion, 3D printing, and connecting peripherals to processors were all highly crucial parts of this project that I had to teach myself.
Looking at the designs that were about a year old, I would not regret saying that they were slightly (to put it politely), ridiculous. But they also show how much the project has matured through time. And in the end, hopefully it would be something that is as ridiculous as possible while still being something doable for a single person in two years.
The biggest change I have made is to forego the moveable center channel enclosure. I do think it is potentially doable, just not at my current level of engineering ability. Another question to ask is: sure it would look cool, but how big of an effect would it actually make?
To begin explaining the new improvements, I feel the need to label each driver so I can refer to them clearly. I had been labeling them as they do in theaters: the left side facing the listener is actually the "right" side.
Right woofers: R1, R2, R4
Right upfiring tweeter: R3
similarly:
Left woofers: L1, L2, L4
Left upfiring tweeter: L3
Center channel woofers: C1
Center channel tweeters: CR1, CR2
Enclosures from right/left to center, RE1, RE2, RE3, CE1, LE3, LE2, LE1
The new center channel enclosure is one solid piece that connect to the L&R of the soundbar. It consists of a single 2.5'' woofer and two 3/4'' tweeters.
The left and right channel of the soundbar each contains 3 2.5'' woofers, and 1 upfiring 3/4'' tweeter. The rms power of the soundbar would be 300w.
Here is another picture that better demonstrates a single enclosure separated from the rest of the body. RE1 is constructed using 3 pieces of 3D printed parts: the faceplate, the right half of RE1 and the left half of it where where it is shaped in a semi sphere. The enclosure is bolted together without glue, therefore, in order to achieve the seal, tolerance in fitting of the pieces must be extremely low for printing standard. The pieces are connected by 4 4mmx40mm bolts. You may think of the bolts almost as a rebar in concrete, the longer it is, the less the enclosure itself flexes. The drivers are then mounted on the faceplate, with two bolts on R1 bolted onto the left, spherical body as shown in the picture. The result is a decently stiff body without using any glue.
On the right half of RE1, a rounded rectangular opening and a much smaller circular opening can be seen. The bigger opening is used to connect to the next section of the enclosure, almost in a Lego manner. The smaller opening allows for the four speaker wires to go through. Anecdotally, I had to lower the wire gauge from 14 to 18. Eventually when all the wires from R1-R4 join together and go through the opening in RE3, 18 gauge is just thin enough for all of them to fit very snuggly through a 1cm hole.
Here are RE1 and RE2 connected together. As you can see, the tweeter is hidden behind the faceplate. The tweeter itself is held on by two pieces bolted together, and that body is then bolted to the faceplate by three bolts as shown here. The tweeter is positioned at a 40 degree angle so that it can more efficiently reflect those high frequencies from the ceiling. The internal wall of the tweeter horn was also carefully designed, though I doubt it has any major effect since the other side is left open in order for it to sit flush with the faceplate.
For now, we have R1, R2, and R3. It must be noted that RE2 would be welded to RE3. On the inside, a 4cm thick pillar goes from the left wall of RE3 all the way to the right wall of RE2, where it is then bolted down. This is also done to achieve stiffness.
A question that one might ask is: why use so many bolts instead of glue? Gluing everything would certainly be much easier than having to design each screw holes and evaluating tolerances, however, if anything goes wrong—for example, one of the speaker wire came loose—the entire enclosure is toast. A simple body such as RE1 with three pieces takes more than a day to print, and it is simply something I cannot afford to lose. For RE1, RE2, and RE3 together, there are, in total, 12 pieces than needed to be printed. That is about 3—4 days of print time, minus the chances of a print failing 10 hours in. My philosophy, therefore, is to make every component as accessible as possible.
The welding could be done here because the faceplates of RE2 and RE3 are bolted, through support pillars, to the back of the enclosures. Therefore, if something goes wrong internally, I can still very easily access the enclosure.
If we double the time it took to print a half of all of the enclosure parts, the approximate time to print both left and right side would be over a week. Realistically, though, it would take much longer: I would need to change out the filaments when they run out, a print might fail from a gust of wind, and I would not be able to stand next to the printer waiting for it to finish all day.
The amount of time it actually took me, if I recall correctly which I rarely do, was about half a month. Then, I moved on to printing to outer case, which was a whole different level of nightmare.
The reason why it was so difficult to print case parts is fairly simple to understand. If a print is only 5mm thick, yet 150mm tall, the contact area it has on the print bed would be ridiculously low. The taller it prints, the higher the chance that the extruder head would knock the skinny print over. It also risked warping off the heat bed because tight corners in close proximities (2 90 degree turns, 5mm apart) trap much more heat than the rest of the printjob. As the print cools down, it shrinks and pulls. As a result, because the tight corners store more heat, they shrink a lot more and eventually pull themselves off the print bed. I have a garbage bag filled with failed attempts of printing outer case parts, there must have been over 50-60 of them, all with minor tweaks from the last one. That bag represents months of my life that I will never get back.
I am proud to say I did not give up, and here it is, in all its glory:
If you ignore the Threat Level Midnight cameo, you will see that the soundbar is not finished at all. The final part would have to be figuring out UX: buttons, display, input selection, etc. Unless that is all designed and confirmed, the center channel enclosure cannot be printed.
Written 9/20/2024
Comments