DIY 2-Way Bookshelf Speakers
Designed by Tony Gee
This is a pair of HATT mkII speakers that were designed by Tony Gee. They use aluminum cone drivers from Seas (19TAF/D tweeter and the L11RCY/P woofer), feature a second order series crossover at 4,500Hz, stand just under 10" tall, 6" wide, and about 8.5" deep, and have a nominal impedance of 4 ohms (a result of the series crossover). Due to variations in construction and final appearance, my pair features a slightly modified positioning of the drivers. Because of the location of the vent and my desire to have a single piece of wood for the finished front baffle, I removed the bottom part of the tweeter's face plate so that I could position it more closely to the woofer than Tony originally did (they are spaced apart by 3.5 inches, center to center - just shy of the theoretical best spacing of 3.0" center-to-center. This is based upon dividing 13568 by the crossover frequency of 4500Hz which indicates the maximum spacing required to avoid half wave beaming/interferences in the crossover region). Also, I used the crossover from the mkII version of the speaker and added a switch on the input cup to be able to engage or defeat the Zoble network on the woofer. The original cabinet and crossover plans are available at Tony's web site linked above.
Listening Impressions: Click here for an updated Evaluation of the a40 amp and HATT combination.
My hat is off to Tony (no pun intended) - this is one very nice sounding little speaker! As described, the bass it produces is very dependent upon exactly how the speaker is placed in the room. After I finished building them, I placed the speakers on the floor and cranked up the volume. I was very impressed by the quantity of bass that they produced! However, after placing them on the wall-hanging shelves that I made (positioned approximately 5 feet above the floor), the bass is much reduced. In my case, this works well, since they are in my home office - located directly beneath my 1 year old daughter's bedroom. I usually listen while working at night so the diminished bass response from being so far off the floor is perfect for me. If I used them in a different environment, I would be more inclined to add a small subwoofer to them... Interestingly, the bass response (to my ears, at least) seems to improve relative to the remaining frequencies when they are driven at higher volume levels. Perhaps they are not yet fully broken in...
Compared to my Atlantic Technology System 350 left and right speakers ($1000 per pair), these little guys sound noticeably more "clean" (I shouldn't really be surprised given Tony's waterfall plot) and produce a little more bass and midrange (the 350LR speakers feature a 3rd order roll-off at 80Hz as per THX specs) giving them an every so slightly "warm" sound which I like very much. They image very well and cast a sound stage that is superior to the 350's. Overall, these speakers do the "disappearing act" better than the 350's do. When connected to my DIY Pass Labs a40 amp (Class A amplifier, see link at top of this page), these speakers sound amazing for their price point!
The crossover networks are constructed using high quality components: Alfa-Core 14ga copper foil inductors, Sonic Craft Sonicap capacitors, and Mills 12w wirewound non-inductive resistors. I have read lots of good things about the Sonicaps and Mills resistors, so chose them based upon the combination of public opinion and their median price point. Total cost for two completed crossovers (including Zoble networks) was approximately $170, just over the cost of the drivers for this project. A number of well respected speaker designers recommend spending similar amounts of money on the drivers and the crossovers to help insure that each part has the opportunity to perform at its best. Each of the components were matched with my LCR meter across the two speakers to help make the final product as closely matched as possible.
I assembled the crossovers on some 1/8" thick 3" wide boards I had left over from a previous project. The components were strapped to the board with plastic wire ties and then glued in place. My plan for the wire ties was simply to hold the components in place while the glue dried, but I decided not to removed them after the crossovers were completed. The crossovers were assembled with Wonder Solder and the inductors were placed at right angles to one another to help minimize interference and interactions.
The drivers are from Seas (in Norway) and include the (H561) 19TAF/D 3/4" aluminum dome tweeter and (H760) L11RCY/P 4.5" aluminum cone woofer. Total cost for two sets of drivers was approximately $150 (I ordered them from the UK, but forget exactly where, its been quite some time...). Since my cabinet construction was a little different than Tony's original plan (both construction methods yield the same cabinet dimensions), I needed to trim a little from the bottom of the face place of the tweeter in order to get the drivers a little closer together. With The entire front baffle on my speaker is a single piece of wood in my cabinet design, whereas the original plans show the front baffle piece ending at the slot vent.
To remove a portion of the tweeter flange, I used 60 grit sand paper placed across the curved top of my sanding block. Where the sanding block has teeth to hold the sand paper in place across its flat bottom surface, I merely turned the sanding block upside down and installed the sand paper such that the grit went across the convex surface of its top. Then, holding the tweeter in the palm of my hand so that the face place of the tweeter was parallel to the length of the sanding block, I proceeded to sand away part of the tweeter flange. The left most picture below is a diagram of how I stood the tweeter on its edge so that the dome of the tweeter was oriented above the side grip of the sanding block and moved the tweeter back in forth motion across the curved top of the sanding block. It takes a pretty firm grip and removes only a little of the flange at a time, but keep moving the tweeter back and forth across contour for the sanding block eventually removes a convex portion of its flange so that it will fit snugly against the curved top of the woofer flange. As you can see, I ended up removing a good deal of the tweeter flange. Be sure to cover the dome of the tweeter with tape (though don't stick the tape to the dome!) to prevent plastic dust from entering the moving mechanism before sanding!
To make the cut-outs for the drivers was a little tricky, since I don't own a router. Instead, I used my drill press along with various drill bits and a 1.5" sanding drum. This took a little longer, but worked out fairly well. First, I placed the drivers on the front baffle and traced out the rear profile for each of them allowing a little more room for the wiring terminals to pass through the baffle. Using a 1" wood boring bit, I made a series of holes that overlapped just slightly until I could knock out the middle piece, leaving a very rough hole for each driver. Then, using the sanding drum, I smoothed out the holes until the drivers easily passed through and fit well against one another. When using a sanding drum, I found a 50 grit paper to work very well - 80 grit just seemed to generate a great deal of smoke after a short while. Another important consideration when using a drill press with a sanding drum in this manner is dust containment. My first effort left me covered in a layer of sawdust about an inch thick! Next time around, I clamped the hose from my shop vac approximately 1/4 of an inch from the back side of the sanding drum and that solved the problem quite nicely!
Next, I placed the drivers on the front baffle and traced the outline of the flanges for both speakers - you can see my tracing in the picture that shows both baffles together. Next, I fired up the trusty sanding drum again (using the highest RPM setting on the drill press) and began making the recesses for the drivers. To do this, you must move the table of the drill down so that you can place the sanding drum in the cutout on the baffle, and then carefully raise the table until the drum will remove just enough material to allow the driver to sit flush against the front of the baffle. This requires a little trial and error - just take a little bit at a time. Sand out the entire perimeter of the driver, remove the baffle to test-fit the driver, and then do it again, taking a little more off to make the recess deeper. I did the same for the woofer, but just went a little bit deeper into the baffle since the woofer flange is thicker. The final product is shown in the picture below on the right. The cutouts are not exactly perfect, but they worked well enough for me. The actual center-to-center distance of the drivers in the completed speaker is about 3.5 inches - I think the only possible way to get any two drivers closer together is to use concentric mounted drivers! The slot was made by drilling a series of 1/2 holes next to one another and then smoothing them out with a 1/2 stock removal bit. Just slide the wood back and forth along a guide until the slot was relatively smooth. I also used the sanding drum to take the edge off the inside of the hole for the woofer so the wood would not interfere with the back wave of the cone.
The cabinets are made from 13-ply 18mm void-free marine grade Baltic Birch plywood ($40 for a 5 foot by 5 foot sheet). This wood is very heavy, dense, and rigid (all highly desirable for speaker enclosures!) and has a nice finish when sanded smooth. I found wood cutting to be a very tedious task for such a small speaker. Because the cabinet is so small and the wood is rigid, there is no "fudge factor" if you don't cut exactly properly! I found an error while cutting as small as 1/2mm to require cutting a new piece of wood. In one place, the final cut on the cabinet left a gap that I had to close with a small shim that I trimmed from the edge of another board. The good news is that the shim blended seamlessly into the cabinet due to the lamination structure of the plywood! The photo below on the left is the cut wood for each speaker prior to assembly. On the right is a picture of the assembly process for the cabinet. I started with three boards to insure that all of the angles in the cabinet would be exactly 90 degrees. This way, I could perform any necessary "shifting" of the parts before the glue dried and the bond became permanent. Just remember to use scraps of wood under your clamps so you don't damage the wood of the cabinet while clamping it together!
The wood panels were glued together using Gorilla Glue, which is an excellent adhesive. It is an expanding glue (so it must be clamped) that fills small voids in your seams, is completely sandable, and creates an air-tight seal. Before gluing the cabinets together, I bonded two pieces of the plywood together at a right angle (forming an "L" shape) and let it dry for 24 hours. Curious to test its strength, I tried to break the bond both by hand and by stomping on the glued joint. Nothing moved! Finally, with the "peak" of the L-shaped wood place up in the air and the two ends resting on the concrete basement floor, it took three extremely hard blows from a 16oz claw hammer to separate the two pieces of wood! The Gorilla glue held, while the layers of the plywood actually separated and gave out first! This is VERY strong glue! Before the cabinets were completely assembled, I mounted the crossover to the top panel of the cabinet, directly behind the tweeter, and placed the Zobel network on top of the board that made the vent. The boards containing the networks were given a liberal coat of the expanding glue and then screwed down to the cabinet. As the glue dried, it oozed out around the thinner board insuring that there won't be any rattles or buzzes because something is moving around inside the cabinet. The photo below on the left also shows where I rounded the inside portion of the top of the vent for the speakers with the sanding drum.
Next, the rest of the cabinet was assembled. Usually, I added one panel to the cabinet each day, allowing the glue to dry for 24 hours before adding the next panel. While it took about 2 weeks for me to assemble the cabinets, it took me only 5 minutes each night to properly glue, fit, and clamp each panel. The photos above show the completed cabinet ready to be closed. Before this final act, I glued some left over carpet scraps to the insides to act as damping and help reduce unwanted resonance - I was unable to source lead bitumen locally. Finally, the cabinets were closed. I was really surprised just how heavy these little guys were before even adding the drivers! The seams of the cabinets came out very nicely. The photos below show the top corners of one cabinet. You can see where the glue oozed out a little while I was claming them. I tried to wipe up the excess as it became obvious, but as the glue dried, it continued to expand. Since the cured glue is easily sandable, it came off very easily and cleanly and didn't leave any stains on the wood. From this perspective, you can also see the recesses that were cut with the sanding drum for each of the drivers. Wrapping your knuckles on the side of the completed cabinet yields only sore knuckles! Not a hint at resonance from the cabinet - this is one heavy and solid little speaker! Next up was one final pass with the sanding drum to round over the edges on the front baffles of each speaker. All you need to do is remember to keep moving as you pass the speaker from one side to the other of the sanding drum - pausing, even slightly, causes the drum to dig extra deep that requires extra sanding by hand to smooth out. Fortunately, I practiced doing this first with some scraps of wood!
Many, many hours of hand sanding later, the cabinets were ready for some stain. The biggest trick was sanding enough to keep the generally size and curvature of the round over on the front panel the same for all edges on both speakers. To prevent dust from getting inside the cabinets, I taped paper into the recesses for the drivers while I sanded. Next up was two coats of Minwax "natural" stain and three coats of polyurethane, lightly sanded in between coats. Then the woofers were broken in with a 10Hz sine wave at high level for approximately 20-30 hours before the I sat down to listen to them.
The final result is pictured above. Pictures at the top of this page show the completed speakers sitting on small wall-mounted shelves that I made using left over scraps of wood. Overall, I am very pleased with both the sound and the appearance of these little guys! Nearly two years in the making (so I'm easily distracted, OK?) these speakers are a pleasure to listen to! Special thanks to Tony Gee, the designer of these little wonders!