Finished speaker, V1.1
Finished Speaker 1 Finished Speaker 2
Click any picture for a larger view.
(construction details below)

January 2008 - updated slides 7, 11, 12, 108, 115
November 2007 - added appendix E: previous projects
October 2007 - added some new 'finished' shots:

Finished 3 Finished 4 Finished 5 Finished 6 Finished 7 Finished 8 Finished 9

(in chronological order)

Photo Number 1 Photo Number 2 Photo Number 3 Photo Number 4 Photo Number 5 Photo Number 6
Photo Number 7 Photo Number 8 Photo Number 9 Photo Number 10 Photo Number 11 Photo Number 12
Photo Number 13 Photo Number 14 Photo Number 15 Photo Number 16 Photo Number 17 Photo Number 18
Photo Number 19 Photo Number 20 Photo Number 21 Photo Number 22 Photo Number 23 Photo Number 24
Photo Number 25 Photo Number 26 Photo Number 27 Photo Number 28 Photo Number 29 Photo Number 30
Photo Number 31 Photo Number 32 Photo Number 33 Photo Number 34 Photo Number 35 Photo Number 36
Photo Number 37 Photo Number 38 Photo Number 39 Photo Number 40 Photo Number 41 Photo Number 42
Photo Number 43 Photo Number 44 Photo Number 45 Photo Number 46 Photo Number 47 Photo Number 48
Photo Number 49 Photo Number 50 Photo Number 51 Photo Number 52 Photo Number 53 Photo Number 54
Photo Number 55 Photo Number 56 Photo Number 57 Photo Number 58 Photo Number 59 Photo Number 60
Photo Number 61 Photo Number 62 Photo Number 63 Photo Number 64 Photo Number 65 Photo Number 66
Photo Number 67 Photo Number 68
Photo Number 69 Photo Number 70 Photo Number 71 Photo Number 72 Photo Number 73 Photo Number 74
Photo Number 75 Photo Number 76 Photo Number 77 Photo Number 78 Photo Number 79 Photo Number 80
Photo Number 81 Photo Number 82 Photo Number 83 Photo Number 84 Photo Number 85 Photo Number 86
Photo Number 87 Photo Number 88 Photo Number 89 Photo Number 90
finalisation + appendixes
Photo Number 91 Photo Number 92 Photo Number 93 Photo Number 94 Photo Number 95 Photo Number 96
Photo Number 97 Photo Number 98 Photo Number 99 Photo Number 100 Photo Number 101 Photo Number 102
Photo Number 103 Photo Number 104 Photo Number 105 Photo Number 106 Photo Number 107 Photo Number 108
Photo Number 109 Photo Number 110 Photo Number 111 Photo Number 112 Photo Number 113 Photo Number 114
Photo Number 115 Photo Number 116 Photo Number 117 Photo Number 118


Email me: fpara AT oricom DOT ca

Created with Web Album Generator

In a nutshell, this speaker design combines world-class drivers and an excellent subwoofer into a civilized volume (120 L), delivering mind-boggling musical detail, and extremely rich, high quality bass, without the need for any signal processing. dayton hi-fi rss315hf diy monster huge speaker curved opera sauvage bosch mercury worklamp polyester hifi design shielded outstanding passive radiator gershman Solen inc. acoustics solen crossovers 20hz home theater subwoofer vented low f3 bracing damping veneer J-roller wood stain wood kote fini-tech 3000 urethane finish maple cedan pan-l-trim mohawk fil-stik xt25sc30-04 xt25tg30-04 sc30-04 tg30-04 high fidelity passive adire pr15 xmax cutout peerless menhir mdf thick layers glue inserts 12db 3-way Sansui au-9500 subwoofer vifa xt-25 xt25 dayton hi-fi rss315hf diy monster huge speaker curved opera sauvage bosch mercury worklamp polyester hifi design shielded outstanding 13w1v2 passive radiator gershman acoustics solen crossovers 20hz home theater subwoofer low f3 bracing damping pr15 xmax cutout peerless menhir mdf thick layers glue inserts 12db 3-way subwoofer Sansui au 9500 Scan-Speak scan speak revelator 7000 R 2904/7000 scan-speak R2904/7000 scan-speak 15W/8530-K01 15W 8530 birch veneer menhir I cedan pan-l-trim Wood Kote oil based jel'd stain Bassbox 6 pro This project was my fourth DIY design, whose goals were the following: 1. To use Hi-Fi components; 2. To give the design a natural competence for extended bass and use a passive crossover, for exquisite sound quality and purity. 3. To come up with a sexier look than a box. It took no less than this unit to pull ahead of Vifa's XT25, Which was the runner up. The level of detail this tweeter delivers is astounding. I chose a 8-ohm unit in the name of tonal balance, to match the bass driver. The custom crossover will make everything compatible. This top-quality midrange sounds so clean and natural, yet so dynamic that it's almost frightening. At only 84.5db @ 1W, it's a perfect match for the inefficient subwoofer. My entire build revolved around this driver, which was chosen over dozens of other 10", 12" and 15" contenders. Simulations revealed it can provide outstanding bass extension in feasible volumes, in the top 3 overall. Runner ups required more volume, didn't go as low, or cost alot more. Modeling software really demystifies common misconceptions about the physics of bass, and can make up for its cost if you consider all the mistakes that can be avoided thru simulations. I modeled hundreds of drivers of all sizes, using three different programs, all of which predicted this flat response for the Dayton RSS315HF in a 120 L box (4.3cuft). Notice the <20hz F3 ! Of course, the passive radiator's effect is taken into account. However I tested all drivers the exact same way, and very few managed to render this linear low-end response. The very last slide (#115 - Appendix D) explains how a 12" driver was chosen for bass over much bigger ones. The amazing Dayton Hi-Fi subwoofer. This 4-ohm, aluminum-cone subwoofer uses a heavy magnet (150oz) and has very low distorsion. Using such a driver in a large three-way system provides unmatched bass extension but has its drawbacks: lesser efficiency. Because most midrange drivers of matching impedance are significantly louder, I chose a 8-ohm midrange to help solve this unbalance. The tweeter also gets its own L-Pad circuit in the crossover. These solutions provided tonal balance, allowing all 3 drivers to blend their performance over a notably wide frequency range. Connectors are nice and tight enough, that there is no need to solder the wires. Using a big passive radiator such as this one can improve a system's bass extension, allowing the same low-end response as a significantly larger ported enclosure. yellow = vented box, lone 12" driver. F3 = 22hz green = 12" driver + 15-inch Passive Radiator; F3 = 19hz = same as a 60% larger box. The acoustical effect of a passive is very similar to a port, but without any of its drawbacks. In simulations, this model was a perfect match for the active driver. This simple driver is essentially a MDF hardboard with suspension. But it uses take up very little volume inside the box, and has no issues with placement, or air speed. It can also undergo numerous calibrations with ease. Looking at countless speaker pictures, my eyes landed on the beautiful 'Opera Sauvage' speaker by Gershman Acoustics. This is the speaker whose looks inspired my project. For me, getting curved sidewalls on such a scale was easiest by trimming down a blocky outline. Based on my experience with 3 previous designs, the average cabinet thickness I aimed for was at least 1.25" for the bass section, at least 1" for the midrange cabinet, and a beefier 2" for the pivotal front baffle. The speaker is fairly large, although it's the smallest I've made. These drawings roughly represent the cuts I had to make. The initial 4x8 sheets each got 4 cuts (the horizontal lines) at the hardware store. This is all the MDF as it came right out of the hardware store. Nearly a quarter-ton! 20 cuts were made on site, to make them fit into my car. All these strips were larger than the end measurement, allowing for some last-minute flexibility. 25 slices, 20 inches thick. These dimensions are the final ones. The numbering system is not the one I used for cutting. I could never have done accurate cuts at home without my reliable BOSCH portable saw. This is about half the total amount of dust and leftovers this operation made. Another view. Good thing I was outside! That dust gets everywhere. This is all the parts (68) after they have all been cut down to their exact sizes. Another view. A few power tools were used in this project. A sidewall gets its final layer glued on. Sidewall, front side up. Notice the 1" thickness extended to meet the front. Another view of the sidewall. On the right, it extends some 5/8" into the rear baffle, which helps make a very strong glue bond. Special holes need to me made to efficiently squeeze the glue between 2 large layers of MDF: Firstoff, sticking the drill bit just far enough inside the drill chuck, or using a spacer if needed, ensures you don't exceed your drilling depth and traverse the double thickness. The sidewalls meet the cabinet's floor, made of two 5/8" layers of MDF tightly glued together, for a total thickness of 1.25". This tool is no router but it does the trick. This circle-making modification was crude, but it worked perfectly. I used an aluminum strip and drilled some holes in it according to the different required cutout sizes. Some circles are the actual cutouts, some are just the edge of the planned inser, and some are made to facillitate overall shaping. After some chisel work, this was the finished inset of the bass cabinet. a 14" circle is cut out. This is the INNER layer. Dusty work! The all-important front baffle needs to have its layers very well squeezed. For the drivers, I always prefer mechanical screws + inserts, although they cost significantly more. Finding the exact sizes and black color was no easy task: I had to go to a specialized merchant. All these screws have allen heads. I carefully plotted the mounting screw hole positions with the drivers in place.. I used a drill press to make the mounting screw holes, ensuring they were perfectly straight. Closeup of bottom face after drilling. Notice the inner board's driver cutout is smaller, for greater strength where it matters. This is after the front baffle was glued on, using these clamps. Driver installation: I must determine the perfect sized-hole for a fit that's tight but not too tight. To avoid reducing the baffle's holding thickness even more for the metal inserts, I devised this system: first I put the driver screws in... Then I screwed in the inserts all the way... These wooden pegs are slices of 1-inch 'broomstick', which I then drilled. I screw them right around the insert. Turning all the way down ends up squeezing some wood glue on the peg's backside, against the baffle! The off-center holes were a blessing in some tighter spots. This woofer is so heavy, the 2-inch front baffle seems like it's just barely thick enough to hold on to it tightly. The three braces were easy to fit in this interior. The distances between them is intentionaly unequal. The rectabgular cutouts were easily done with a jigsaw. The midbox's layered sidewalls, whose top and bottom feature the only angle cuts at 10, are glued to the backside, which is still a rectangle for the moment. The top front. The notches are for the connectors that stick out of the tweeter. Apply glue on the insert... Although a bit more should be applied than what you see on this picture. And voilą! Best results require a tight-fitting insert. Both top drivers share the same inserts. My previous project was thus huge worklamp, which provided abundant mercury lighting for this project. The sealed midbox's face is glued on. This enclosure still needs a rear baffle, and then it goes on to sanding, after which the top will be permanently affixed to the bottom section. The largest MDF pieces were cut out with a jigsaw before the heavy sanding began. Whether going at it 3hz or 300hz, this orbital jigsaw cut thru 1" MDF like it was soft butter. The rear baffle is made of two layers of 5/8" MDF, which are glued together. Here again, thick screws tighten the centre of the baffle whereas clamps can do the job on the edges. Glueing the rear baffle required no less than 2/3cup of wood glue. Some of that glue was applied on the baffle itself, and in respect to the extra surfaces which will meet its extremities. Let's see... large dumbells, neon sign transformers, subwoofers... all heavy stuff baby! Total weight on the baffle: 250 pounds. By the time I realized then a porter-cable planer, equipped with a spiral carbide cutter, was *THE* best way to make a rough cut of this shape, I had already been using a slow but sure belt sander (on the right), with an industrial-grade, 50-grit belt. The former approach would have reduced the dust amounts tenfold, not to mention the duration of this job. I had to empty the Shop-Vac's bucket 14 times. The amount of dust called for a shovel. On the upside, it's very hard to take away too much this way. Amazingly, I used a single sandpaper belt for chewing down alkl four shapes, that's over 7 hours of sanding, and the belt was still as coarse as a new one. Industrial grade baby! Both separate cabinets are superposed for a quick look. This top will also become the midrange cabinet's floor. These bare copper wires cross the 1.25" wall, down into the bottom section, where a similar connector is also present. this damping material converts soundwaves into low-level heat. I used three of these 39"x39" rolls. This adhesive was used to secure the black tiles. It proved to be VERY effective. The front section of the midrange cabinet gets damping too. Its degisn volume ended up being too large for the midrange driver so the back area will be filled with airtight insulating foam. The damping material has been applied on the inside walls. This operation required to cutout no less then 70 pieces of the stuff! Both cabinets are now joined. Besides wood glue, four 3-inch screws were used to secure the top. Veneering these speakers requires very flat and symmetrical sidewalls. Despite looking symmetrical, the current sidewalls have random imperfections, which at most span several millimeters. I devised a simple but effective method to patch this. First, I need two identical 1-inch MDF boards. They are to become templates. I lay the boards down on the speaker face. The only solid reference for this alignment is the center line. Then I trace the sidewall's outline on the templates, using a pencil. Then I cut the outlined boards using a jigsaw. This picture shows the superposed left and right templates. The imperfections are obvious. In order to get identical sidewalls, I will give these templates a curve which fits somewhat within these two dissimilar ones. This apparatus cost little and worked quite well, giving my belt sander an edge for making nice curves at right angles. After joining both templates together, I trimmed them down and made sure to go a little beyond the smallest. The two resulting templates were smooth ans absolutely identical. The arrow shows the pencil line, which is the target profile. Clamps held the template set on one sidewall at a time. This custom-made support features some skids, which are made to lay only on the templates, allowing me to slide my trimmer over the surface and remove only and exactly what's needed. This picture was taken after it's taken a beating. After some trimming, the sidewall's surface had become like the template's profile. A light sanding will make the sidewalls smooth again. After the last step, the surface's imperfections had less than 1 mm. As far as I can testify, MDF boards are much flatter than any other wood. So I cut a 1/inch strip of MDF as a guide, sliding it slowly along the sidewall. I troweled some wood putty where necessary. After sanding I had near perfect flatness. Time to lay the veneer. This is the paper-back type. It's not brittle, and doesnt let glues bleed through. Its 96-inch length means I must leave a demarcation on my 116-inch sidewalls. To the bottom it shall go, for easier concealment. It took no less than a half-gallon of contact cement for all surfaces concerned bu the 2 sidewalls. These strips were positioned to get one inch of veneer overhang in the front. I used the waxpaper method, sliding slowly from under the positioned veneer. Using the J-roller from the center toward the edges was the way to go, about 6 inches of veneer at a time. The bottom section is applied. I trimmed all the veneer with a sharp knife. Pre-glued Iron-on Veneer is absolutely the best, easiest method for flat surfaces. But back in 2006, I hadn't discovered this product, so my next best choice was yellow glue. This is where the project started being 'funner'. Applying carpenter's glue about 10 inches at a time kept to a minimum the surface requiring attention. As a previous section had dried, I lifted the partially-glued veneer and stuck a thin, glue-filled brush along the rift. It is paramount to use light amounts of glue otherwise wrinkles may appear. To apply glue on the remainder of the surface, I used this small roller. This method was so clean and effective, and used such a perfect amount of glue, that I did not even need to use masking tape on the sidewalls. There were no drips. I learned just how invaluable a J-roller was for this job. Pressing very firmly is a must, especially where I used the brush. The pressure weight is 200 pounds. With the forementioned precautions, this yellow glue worked perfectly. I purposely left the last few thousanths of edge trimming for the sander. Depending in its inertia, a passive radiator behaves the same way as a given volume of air in a port. I used modeling clay to calibrate the radiator to the volume of an ideal port. This scale was used to weigh the clay. The clay goes inside a tube that would have been the voice coil structure un an active driver. This quality polyester filler was used as a second damping system and proved to be surprisingly effective in reducing unwanted resonances. I used 5 bags. All the crossover parts and most speaker parts were purchased from SOLEN inc, in Canada, a large distributor of quality speaker components. The onsite technician was helpful beyond words. He was patient, polite, extremely competent, and all his prices are quite competitive. This is my second speaker project with Solen's help, and I was very satisfied with the first AND this one. The crossover parts were fun to sort out. The finished passive crossover after I assembled/soldered it. It's a 3-way 2nd order all-pass which cuts 12db/oct @ 250hz and 2500hz. I'll admit it's not the prettiest thing in the world... but the quality parts are all so very "overkill"! The result exceeded my expectations: the sound definition is quite remarkable. The big crossover fits nicely at the bottom. This project required no less then 35 feet of 12-gauge wire. In June 2007, despite the prospects of an unbearable downtime, I decided to make several modifications to a previously finished speaker (V1.0), which rolled me back to this stage. This slide picks things up from that point. Countless hours later, they were once more primed and ready. Last minute correction on new birch veneer. The oil based "Jel'd" stain is very forgiving and easy to apply. You can't really apply too much, the wood drinks what it can depending on the sanding, and the balance is wiped off. The color is close to 'cream-coffee'. To conceal the single veneer demarcation on the sidewall, I used three different Mohawk wax retouch crayons and smudged the surface to match the wood grain. It's a minor improvement, yet I love those spring connectors! I used four coats of glossy urethane water-based varnish. Folding a 19-inch wide strip of veneer as tightly as this at the top, in such a way as to avoid air bubbles, was one hellish and unforgiving job, requiring hundreds of pounds of pressure per square-inch. Finally, I added a grille to protect the bottom driver. The first sound tests were quite favorable, with the unprocessed bass audible down to 16hz, even at conversation volume. Although not the best bang for the buck, these drivers do crown the design very well. Driving these speakers effortlessly requires some muscle, so I got my hands on this beast: a 1973 Sansui AU-9500, one of the first solid state amps. This thing is built like a tank ! Made entirely of metal and featuring a huge power supply, it weighs-in at nearly 53 pounds. The sound is extremely warm, with abundant bass and rich 'tube-like' sound. It still sounds as good as new, even after 35 years of service. I can testify that its golden reputation is well deserved. Truly an overbuilt gem from Sansui's best years ! Its 4-ohm rating is an understated 120WRMS/ch. In a nutshell, this design achieved my goals:-Spectacular bass! Fast and effortless, with a level of presence very uncommon in 3-way designs, all from a 100% flat amplifier; -F3 below 20hz without any signal correction -Extremely detailed rendering in the upper range, low distorsion; -Better looking than my previous projects. All of these products provided outstanding results. Fini-Tech 3000 water based urethane gloss finish, CEDAN Pan-L-Trim paper-backed veneer, Lepage wood fillers, Wood Kote oil based Jel'd Stain, Mohawk Fil-Stik, Trade Secret light Fill Sticks. Why didn't I use a bigger driver? When it comes to preserving bass extension in ported/PR systems, 12" drivers seem to be the sweet spot for reaching slide #7's hi-fidelity response, within household-friendly enclosures. Hundreds of simulations demonstrated that any other driver size (of 4 or 8 W) will either: A) rolloff well above 20hz, or B) require a huge box. When mounted in their correct enclosure sizes, BIGGER woofers will naturally deliver more loudness AND lower frequencies than smaller woofers (of the same product line). But those sizes quickly get huge for drivers above 12" : 6-10 cuft for 15" drivers, and 10-20 cuft for 18"s. Moreover, such big volumes require boxes with extreme sturdiness otherwise one gets boomy underdamped bass. This is what happens when the box isn't big enough: the same 18" driver, when mounted in say, a minuscule 5cuft box, still averages higher loudness, but the bass extension is gone ! The same thing occurs to a lesser extent, for 15" drivers. This means, every compact subwoofer with a large driver, either uses or needs signal correction (EQ) to produce a flat response. But, is this really suitable for a HiFi system? Type: 3-way, with 12" sub & 15" passive radiator; Impedance: 4 ohms; Sensitivity: 89db @ 2.83V Frequency response: 19hz - 40000hz (+/- 3db); Crossover frequencies (uses Solen caps): 250hz and 2500hz (12db/oct); Damping: Bitumen tiles and polyester foam Internal volumes: Bass chamber: 120 litres / 4.2 cuft; Midrange chamber: 15 litres / 0.5cuft Baffle thickness: Front = 2" ; Elsewhere(bass) = 1.25" ; Top = avg. 1.1" Measurements: Total height = 52.75" at widest = 18" depth = 19.25" Weight: 182 pounds each; distribution is: 105-bass cabinet; 24-midrange cabinet; 10-Damping; 8-Crossover; 30-Speakers; 5-Veneer, screws, etc. Build time: 120 hours Builder: 37yo male, from Canada. My first speaker project. Essentially a new bigger enclosure for someone else's design. Radio-Shack 40-8424 woofer. Indian tweeter. Overall bad upper range drivers and lack of definition. 1.5" MDF. Heavy. 6db crossover. Painted sealed box. These 230-pound, 48" monsters manage to make even this 42" TV look smallish. They consisted of a three-way bass-reflex using a pre-made crossover. The ultra-dead enclosures used 2-inch MDF all over. A 12" peerless 831857 in the 5.5 cuft inner volume gave these beasts house-crumbling bass. Third project. First time assembling a custom Solen Crossover. What a difference in rendering detail ! These stood 48" tall, 5.5 cuft, 185 lbs each, 1.5" MDF all over. 94db. Using 2 peerless 850146 in parallel. Vifa D27TG tweeter, Peerless 850119 mid in a sealed chamber. The bass was rich and colossal, although not quite as much as my latest design.